《Hummingbird Ventures：解码生物快照2023：下一代生物科技公司研究报告（英文版）（115页）.pdf》由会员分享，可在线阅读，更多相关《Hummingbird Ventures：解码生物快照2023：下一代生物科技公司研究报告（英文版）（115页）.pdf（115页珍藏版）》请在三个皮匠报告上搜索。

1、SNAPSHOTDECODING BIO2023The next generationof bio companiesTABLE OF CONTENTSDISCLAIMERAUTHORSDRIVING FORCESEXTREME BIOLOGYINFRASTRUCTURE FOR BIOAI-DRIVEN DESIGNTARGET DISCOVERY&VALIDATIONSCREENINGSPATIAL BIOLOGYNEW MODALITIESDELIVERYBIOMANUFACTURINGCLOSING WORDS2358698106114Designed by wu

2、nderdogs.coDISCLAIMERThe opinions expressed by the authors of this project do not necessarily reflect the views or positions of their respective employers.This project is a community-driven endeavor and the companies listed have filled in their own templates*.The list of companies presented is not e

3、xhaustive and may not include all relevant entities.It should be acknowledged that the companies listed may be at various stages of development,but all are no longer operating in stealth mode.*with the exception of Prime Medicine2Decoding Bio Snapshot2023AUTHORSPABLO LUBROTHSHELBY NEWSADAMEE KAPADIA

4、PATRICK MALONEDAVID LIANDREW PANNUELLIOT HERSHBERGJESSE JOHNSONLUIS VOLOCHWUNDERDOGSKKH ADVISORSDecoding BioHummingbird VenturesBioscene ThoughtsCompoundDecoding BioCantos VenturesDecoding BioKdT VenturesMeliora TherapeuticsImmunaiCentury of BioNot BoringScaling BiotechSerna BioCo-Founder of Immunai

5、Stanford UniversityDesign partnerMedia partner3Decoding Bio Snapshot2023The purpose of this Snapshot is two-fold:Release Date:07-05-2023One of the biggest strengths of the intersection of technology and biology today is the community forming around it.Some call it techbio,others call it computationa

6、l bio,some just see it as biotech.Regardless of the nomenclature,the energy and talent entering this era of biology is special.Embracing that spirit,this project is a community-led effort.We are honored to feature some of the worlds most innovative companies working towards solving hard problems in

7、human health.This is by no means an exhaustive list nor is it one that was cherry-picked.All the companies featured opted-in to the project.Our objective was to feature a wide range of businesses to highlight a diverse set of valuable lessons learned from building in biofrom how teams are structured

8、 to flavors of business model innovation and beyond.We hope you enjoy this first edition of the Decoding Bio Snapshot and leave feeling as inspired as we are.To shine a light on how these new classes of bio companies are organized,how they commercialize their technology,and how they use the tailwind

9、s of automation to improve their performance and achieve scale.There is no shortage of challenges facing the world of biology today but the forcestechnological,cultural,scientificcoming together make it hard not to be optimistic.Happy building,Pablo and AmeeTo map what we believe are pertinent theme

10、s in bio and how these themes are being shaped by changes in culture,improvements in artificial intelligence and machine learning,adoption of robotics,and ultimately better reading and writing of lifes source code.124Decoding Bio Snapshot20231.DRIVING FORCES5Theres no doubt that something massive is

11、 happening within biotech today.Technological advancements paired with a democratization of knowledge and talent is unlocking a speed of discovery previously thought to be impossible.And its not just the advances in biological science that are pushing the industry forward,but those in artificial int

12、elligence,robotics,computer science,and physics that are coming together to allow us to accelerate and deepen our understanding of life and how to engineer it.In the last year,weve seen life be created without an egg,sperm,or womb,all through self-assembly of stem cells.Weve seen large language mode

13、ls like GPT broaden our scope on what human-in-the-loop systems might look like.AI-powered protein predictions grew stronger and gene-editing became more accurate.There were also several therapies approved including GBTs voxelotor treatment for sickle cell anemia,the first therapy to modulate the af

14、finity of hemoglobin/oxygen.This year we also saw important discourse start to form around preclinical models and regulatory process reform.Questions are turning away from speed and access of technology and turning towards the next set of limiting factorsspeed and access of clinical trials.And while

15、 not without its challenges,there are undoubtedly a few driving factors at play behind this new revolution in biology:DRIVING FORCES1Accessibility of data:Costs of sequencing have been slashed,making genomic knowledge more accessible and widely-used.The figure below is the go-to image for showing th

16、e scale at which sequencing costs have come down,from an astronomical$2B to under$1000 today,a rate which far outpaces Moores Law.Companies like Ultima Genomics,Illumina,10X,Pacific Biosciences,and more are developing new technologies for sequencing,such as long-read methods that make understanding

17、complex sequences easier.They are also providing competition to each other in a previously monopolized market.We appreciate the complexity of biological systems and increasingly look to combine genomic data with other-omics to better understand the relationship between systems to inform the underlyi

18、ng mechanism of disease.Applications in clinical monitoring,diagnostics,and care will also be unlocked as technologies like liquid biopsy,point of care sequencing,and embryo-testing come to fruition.Coupled with sequencing(“reading”)are advances in synthesis(“writing).DNA and RNA synthesis is becomi

19、ng quicker and cheaper,allowing us to apply the knowledge gained from sequencing directly.This allows for rapid creation of new cell lines,testing tools,new moleculesexpanding the suite of modalities available,and extends beyond just the healthcare applications of biology.It is not unreasonable to p

20、roject genome sequencing costs to float below$100/genome within the next five years.As the scale and ease of sequencing grows,so too will its impact on genetics and clinical care.The proliferation of sequencing data,especially multimodal and longitudinal data,sheds insight into complex diseases that

21、 span many organ systems.6Decoding Bio Snapshot202334Scale and automation:One of the major limiting factors within the biotech industry is scale.Manufacturing living systems and supporting infrastructure has historically been difficult to parallelize given the required level of control,precision,and

22、 safety.Further,a big part of biological discovery is still trial and error.It is unlikely that we ever fully eliminate the need for physical experimentation and thus running assays and tests in parallel is a fundamental requirement for efficiency.To meet this demand,there have been several advances

23、 in parallelization and manufacturing with new iterations of bioreactors and advances in synthetic biology allowing living systems to be programmed in much the same way we program software systems.The new intersection of science and technology begs the need for better software designed specifically

24、for biotechsinfrastructure ranging from electronic lab notebooks to experiment scheduling to data analysis and bioinformatics.And with this increase in organization comes a need for modern automation platforms,be it for liquid handling(pipetting and running assays),proliferating stem cells,or creati

25、ng new compounds at scale.The focus on interdisciplinary teams and creating a new“tech stack”for modern biotech companies is accelerating this transition.Culture shift:One of the most underappreciated momentum shifts in bio has more to do with culture than with technology.Starting from conception an

26、d following the steps of many of the current pharmaceutical giants,many of todays biotech ventures are entirely founder-led,shifting away from the venture creation models that have come to define biotech.Founding teams often include scientist turned entrepreneurs and have both computational and R&D

27、experts.These companies are often backed by tech(or generalist)investors that are increasingly interested in bio.Modern biotechs also have almost two-times as many computational team members(in proportion to the total R&D staff)when compared with the average biotech.This huge focus on software and d

28、ata infrastructure is a defining factor of their R&D culture,which many people find reminiscent of Silicon Valley“tech”culture.Many new bio companies have flatter organizational structures and employ“tech”product management principles,where the multidisciplinary teams are led by a product manager(wh

29、ich can be a scientist,but leads with business goals and broader context).Together,these changes reflect the influence of technology filtering into bio.7Decoding Bio Snapshot20232Artificial intelligence and machine learning:One of the most notable forces driving biological discovery today is machine

30、 learning.AI/ML is the only way to understand the ever-growing complexity in biological data and to reconcile it with our desire to engineer living systems precisely.In other words,while we may not understand each detail of how biological systems work,AI is the language that can“learn the rules”and

31、allow us to“speak”biology in much the same way mathematics is the language through which we speak physics.This is particularly the case with language models.By training models to learn the relationship between protein sequence and structure,advances in ML have given us the ability to design totally

32、novel proteins with unexpected ease and accuracy.In much the same way models like AlphaFold have transformed protein design,we are starting to see similar promise for small molecules,RNA therapeutics,and new modality design.While its inaccurate to claim AI is a magic system where some desired output

33、 is delivered with full clinical safety and efficacy,it is a tool that is undoubtedly expanding the scope and scale of biological discovery today.Going back to the first point,the increase in data is making applications of AI within biology possible.2.1.EXTREME BIOLOGY8EXTREME BIOLOGY9One of the mos

34、t unique aspects of biotechnology is that we discover our tools rather than invent them.Tools like GFP for biological imaging,Taq polymerase for PCR,and even CRISPR for genome editing,were all discovered in living systems.Fueled by this core observation,a new generation of scientists and engineers a

35、re scaling our search of the natural world for new tools,medicines,and bioproducts.Historically,it has required serious investment to begin research in previously unstudied organisms.Researchers organized into“model organism”communities to share resources and makeprogress.Without modern DNA sequenci

36、ng,it was only possible to study microbes and viruses that could be cultured in the labwhich was a very small subset of Nature.The process of natural product discovery in the pharmaceutical industry was crude and inefficient.Now,we are living through a unique moment where our toolkit for studying Na

37、turefrom DNA sequencing,to DNA synthesis,advanced microbial culture systems,portable gene-editing tools,mass spectrometry,microscopy,and computationis making it possible to renew a much broader exploration of the tree of life.This has been referred to as Extreme Biology.An essay from Arcadia Science

38、 captures this beautifully,saying“we are at an incredible moment in scientific history when an array of distinct but complementary technologies for accessing and exploring diverse species are rapidly maturing,including genome sequencing,gene editing,microscopy,mass spectrometry,phylogenomics,machine

39、 learning,and many,many more.”We are seeing the birth of a new generation of therapeutics companies with an emphasis on finding the solutions that Nature has already inventedacross plants,microbes,and animalsby harnessing the tools of genomics,synthetic biology,and computation.This logic extends acr

40、oss the rest of the growing Bioeconomy,where labs and companies are leveraging novel organisms to move beyond developing drop-in replacements for existing products.The goal is to cultivate evolutionary technologies that extend beyond what we can currently engineer.Decoding Bio Snapshot2023Why it mat

41、ters:Whats changed:How new Extreme Biology companies are different:BASECAMP RESEARCHA new generation of protein design powered by the first high-resolution map of Earths genetic biodiversityFounding year:2019Founders:Glen Gowers and Oliver VinceNumber of employees:22Location:London,UKLast funding ro

42、und:Series ATotal funding raised:$25MWe design high performance proteins for any application(without needing a lab)using our unique,proprietary,map of our planets genetic biodiversity,sourced by our team from extreme environments all over the world.Our platform works across markets,whether developin

43、g proteins for pharma,therapeutics,food,ag,or chemicals.We partner with customers to develop novel IP(new proteins/biochemistries)for their applications.Weve seen most traction in pharma(biocatalysis),therapeutics(gene editing),and nutrition(novel food ingredients).We can annotate complex functions

44、of proteins in silico irrespective of sequence or structural similarity(gold standard today).We can do this because our proprietary knowledge graph adds hundreds of contextual metadata points on top of every protein.By complex function we mean those functions and properties for which there are no an

45、notations(e.g.taste,texture,multi-variate performance etc).This lets our partners develop more predictable products,faster.Our core technology is BaseGraph,our proprietary knowledge graph of biodiversity,which enables us to discover and design novel proteins for virtually any application 10-100X fas

46、ter.Our data is stored in a knowledge graph,a network that captures and recreates the full complexity of the microbial web of life in a supercomputer.Our unique data collection gives us access to unique capabilities;mapping the full genomic,evolutionary and environmental context of each protein,path

47、way and organism allows us to design performance-ready proteins that others simply cant access.Our biodiversity team have built partnerships with some of our planets greatest biodiversity hotspots(22 countries from the Arctic to Antarctica),built on the principles of access and benefit sharing.We li

48、terally go to the most biodiverse and therefore often extreme environments to samples,which gives us unprecedented access to undiscovered biodiversity.In raw size,our database is many times larger than Uniprot,the current gold standard for AI-guided protein design.We are also growing 10 x faster tha

49、n Uniprot.PRODUCT PIPELINE10TECHNOLOGYDecoding Bio Snapshot2023Market:We are focusing on pharmaceutical biocatalysis,therapeutic gene editing systems,and nutrition as our key market areas.These are markets for which we have a unique and unattainable advantage solving a lack of protein diversity for

50、complex substrates(biocatalysis),a lack of freedom to operate(gene editing systems),and complex product functions(novel nutrition ingredients display complex functions that you currently cant annotate in silico,e.g.taste,texture,sweetness).Our vision is a world where biology impacts all aspects of s

51、ociety,from the materials we use,the food we eat,and the drugs we take.Basecamp Research will be the backbone discovery and development engine for this future,and in doing so creating a long-term economic value chain for intact biodiversity.Business model:Basecamp is on a mission to build a bridge b

52、etween biodiversity and biotechnology.This unique access to and understanding of our planets genetic biodiversity is giving us a revolutionary ability to dramatically reduce the need for lab-based protein engineering.We are a platform company that is partnering with the best players in each sector o

53、f biotech.-Proven ability to replace lab-based protein engineering with fully in silico design-tackling previously impossible challenges across pharma and nutrition-A unique map of natures biochemical interactions,relationships and interdependencies that is many times larger than Uniprot(the current

54、 gold-standard database)allows us to achieve AI-based design results that are simply unavailable to those using Uniprot.-Biodiversity partnerships in over 20 countries from the Arctic to the Antarctic,plus more than 75 expeditions in the last 2 yearsCOMMERCIALVALIDATION11Where Uniprot is a poor data

55、set for AI-based design,BaseGraph allows us to see full genomic,evolutionary and environmental context,giving us the ability to custom-design high performance proteins like never before.Why is this a breakthrough?We remove the need for protein engineering.Our ability to annotate the complex function

56、s,properties and performances of proteins and therefore design and map a to any reaction or end application has allowed us to effectively replace lab-based protein engineering.Decoding Bio Snapshot2023Oliver Vince,PhDGlen Gowers,PhDPhil Lorenz,PhDSybil Wong,PhDCo-FounderCo-FounderCTOHead of Partners

57、hipLeadership TeamENVEDA BIOSCIENCESTo engineer new drugs from nature.Founding year:2019Founders:Viswa Colluru,Ph.D.Number of employees:175Location:Boulder,CO USA&Hyderabad,IndiaLast funding round:Series B1Total funding raised:$175MOur most advanced assets:1.A novel oral and topical non-opioid analg

58、esic candidate with unique pharmacology to address pain and itch.2.A first in class candidate that modulates neutrophil biology for a range of chronic inflammatory diseases3.First-in-class candidates that modulate the NLRP3 inflammasome pathway in different tissue compartments.These assets were sele

59、cted due to high Platform-therapeutic area fit,medical and commercial need,and feasibility of clinical development.We expect multiple candidates to enter IND-enabling in 2023 and Phase 1 dosing in 2024,after which we will have more data.Our platform allows natural product drug discovery at scale thr

60、ough the annotation of(i)structure,and(ii)function of completely novel chemistry directly in pools,i.e.,without requiring isolation of individual compounds.We annotate structure with high-accuracy through the use of mass spectral fingerprints interpreted by proprietary machine learning models.These

61、models predict 2D chemical structure directly from the mass spectra without the need of a reference database.We annotate function through a variety of high-throughput screening platforms that rapidly pinpoint active and inactive molecules from compound pools tested in phenotypic,target-binding,and i

62、nteraction assays.Natural products simultaneously represent the most validated and yet untapped opportunity in drug discovery:50%of approved small molecules come from the 5%of the natural world.Technical bottlenecks in searching the other 95%has led to its exclusion from modern drug discovery workfl

63、ows.Our platform removes historical bottlenecks to bring natural product drug discovery to the 21st century.PRODUCT PIPELINE12TECHNOLOGYDecoding Bio Snapshot2023Markets:We are currently targeting inflammation,fibrosis,and neurosensory biology within the dermatology,GI,and pulmonary therapeutic areas

64、.Apart from being large markets with growing medical need and enabling facile clinical development through PoC,these diseases are highly consistent with the historical use of medicinal plants.Envedas 10 year vision is to deliver the first high-resolution chemical map of the world.As we unlock the wo

65、rlds chemical code,we will create and systematically harvest opportunities across industries as diverse as pharmaceuticals,food&agriculture,manufacturing,diagnostics,and forensics.Our primary business model is to collaborate with strategic pharma partners to maximize the value of our pipeline assets

66、 and access complementary expertise,and also partner in diseases outside our core therapeutic areas,such as neuroinflammation and oncology.COMMERCIAL1313Our platform enables the identification of drug leads from natures dark chemical space at scale,(i)delivering new mechanisms of action that lead to

67、 discovery of new biology,and(ii)targeting proteins widely thought of as undruggable.We do this through the rapid prioritization of tractable chemistry from millions of novel chemicals found in nature,beginning with terrestrial medicinal plants.Without our platform,identification of novel bioactive

68、compounds from nature would involve isolation of individual metabolites,their individual structural characterization by NMR,and individual testing in bioactivity assays.On average,dozens of such molecules would need to be isolated prior to finding a single lead-like molecule amenable to medicinal ch

69、emistry optimization.We conservatively estimate that our platform is 1000X faster than this traditional workflow.We believe that our approach will ultimately deliver greater success in clinical development than the industry average.Our belief is driven by(i)historical success rates of natural produc

70、ts,and(ii)the relevance of human priors for the identification of high-potential chemical space in medicinal plants.At the laboratory level,our platform functions like a tag-less pooled screening technology.This allows us to characterize protein targets,biological activity,and even organ distributio

71、n of known or novel compound pools at scale.We are using this to build the worlds most chemically diverse biologically annotated dataset.We have validated our platform by several gold-standard measures of annotating structure and function of pooled compounds.For example,we recently published benchma

72、rking of high-accuracy drug-likeness property predictions(MS2Prop)of novel compounds based entirely on mass spectra alone.We have also recovered dozens of known active compounds from literature at a fraction of the cost and time(not disclosed).Decoding Bio Snapshot2023Hannah GordonDan WeeAugust Alle

73、nViswa ColluruMark DeegJoe RokickiSotirios KaranthanasisVanitha SekarVice President,ProductChief of StaffChief Platform OfficerFounder CEOChief Medical OfficerChief Technology OfficerChief Science OfficerChief Business OfficerLeadership TeamFAUNA BIOImprove human health and lifespan by scalably mini

74、ng extreme biology for new therapeuticsFounding year:2018Founders:Ashley Zehnder,Linda Goodman,Katie GrabekNumber of employees:17Location:Emeryville,CA,USALast funding round:SeedTotal funding raised:$13.2MWere creating a pipeline of novel genetic targets and molecules that have the potential to impr

75、ove treatment paradigms for many diseases of high unmet need.Faun1003 is a small molecule identified through our proprietary LEO platform that has demonstrated in-vivo efficacy in animal models of pulmonary fibrosis and pulmonary hypertension.This program is in late lead optimization,represents a tr

76、uly novel mechanism addressing underserved patient populations and is targeting clinical entry in early 2025.We have eight discovery programs in various stage of development.Our most mature programs are in pulmonary fibrosis,retinitis pigmentosa and cardiac fibrosis with early discovery programs in

77、inflammatory bowel disease,tauopathies and renal disease.The combined TAM of these initial programs is over 60B.The combined TAM of these initial programs is over 150B.We have had the human genome sequenced for more than 20 years and the top 10 disease killers have barely changed.76%of new drug appr

78、ovals in the past 2 years are designed against known targets or pathways.We need truly novel datasets and approaches to discover new treatments for diseases of high unmet need.Over thousands of years weve successfully tapped into nature to identify new medicinal compounds,we have no meaningful or sc

79、alable approaches for tapping into natural animal models of disease resistance to identify new targets and pathways.Extreme biology and disease resistance have been the source for some of our most successful and innovative drugs(GLP1 agonists,ACE inhibitors,PCSK9 inhibitors),but the current methods

80、of discovery are stochastic and unpredictable.PRODUCT PIPELINETECHNOLOGY14Decoding Bio Snapshot2023Current market and rationale:Our current market is focused on specific subpopulations of pulmonary fibrosis,representing rare disease markets with high unmet need and no effective treatments.We are sta

81、rting with rare and orphan disease markets to validate our novel targets and compounds to rapidly de-risk programs.As our discovery process is anchored in genomics,we can identify rare disease populations with alterations in key genetic pathways through our Convergence TM platform.Subsequent program

82、s(and partnered programs)will focus on more complex diseases of high unmet need(including obesity and heart failure).We envision a world where we can learn from nature how to treat our most deadly diseases.Ten years from now we anticipate seeing our own drugs improving patient lives and tackling mul

83、tiple diseases of aging through mechanisms discovered in extreme biology and disease resistance while having a robust discovery pipeline built on a network of collaborations to access new species.Our current model is internal asset development with a focus on rare and orphan disease populations whil

84、e de-risking discovery programs in complex disorders through pharma partnerships that generate early stage revenue,developmental milestones and royalties.We have one disclosed early-stage partnership with Novo Nordisk focused on obesity which leverages insights from the 13-lined ground squirrel(a hi

85、bernation model with extreme metabolic plasticity).We are interested in one additional partnership with a pharma company that brings expertise in a complex disease area where we have unique biological insights through our proprietary animal model data.We have 4 academic partnerships providing us acc

86、ess to additional species models.COMMERCIAL15Fauna Bios Convergence TM platform(in the image above)mines extreme biology and resistance across hundreds of species,linking traits directly to human disease and predicts compounds based on expression data.This process streamlines the process of going fr

87、om genes to drugs and unlocks over 150 new species for discovery.We are the first and only company to be able to directly link extreme biological traits in mammals to human diseases in a way that is scaleable and high throughput.Drugs with genetic support across multiple species are more likely to f

88、unction in humans.Our proprietary knowledge graph,Centaur,rapidly matches extreme animal traits to known human disease biology and Leo predicts new small molecule entities that can match expression signatures from disease resistance traits in animals.We have in-vivo efficacy data in 4 diverse diseas

89、e areas(pulmonary fibrosis,neuronal metabolism(focused on retinal indications),obesity and cardiac damage)and additional novel genetic hits from our programs in neurodegeneration and inflammatory bowel disease.Decoding Bio Snapshot2023Ashley ZehnderRui SongPhil McNamaraKacey HaptonstallEvelyn TranKa

90、thleen KeoughPoornima NeelaDeek GurugeBen SajdakBryan BurkeyKatie GrabekLinda GoodmanJacqueline GaleasFounder and CEOSenior Research ScientistBioinformatics Platform DeveloperTranslational Operations ManagerSenior Research ScientistSenior Scientist,Computational BiologyResearch ScientistBioinformati

91、cs ScientistDirector of Emerging Animal ModelsHead,Therapeutics DiscoveryFounder and CSOFounder and CTOResearch AssistantThe TeamSHIRUDecarbonizing ingredients used in food and everyday productsFounding year:2019Founders:Jasmin Hume PhD Number of employees:29Location:Alameda,CA USALast funding round

92、:Series ATotal funding raised:$20.5MShirus first branded ingredient,OleoPro TM,is ready for immediate use as a butter,lard,or tropical oils substitute in food.It provides superior performance in application for plant-based meats,as well as a drastic reduction in saturated fats compared to what it re

93、places.Sustainably produced OleoPro TM avoids the environmentally harmful deforestation associated with palm oil,which it can replace on a 1:1 basis.Shiru has additional commercial traction across several functional areas,most notably a replacement for egg in baked goods and a replacement for the ch

94、emical binding agent methylcellulose.There are several other high value,functional ingredients in the pipeline at Shiru that range from natural bioactives to functional emulsifiers.currently under development.Shirus proprietary platform,FlourishTM,is a first-of-its-kind platform that directly links

95、protein sequences to functionality.FlourishTM is a full stack technology solution that spans AI-driven digital discovery and pilot scale protein production.This full end-to-end tech stack enables Shiru to go from digital discovery to end-use testing in a matter of weeks.AI is at the heart of Shirus

96、discovery capabilities.The discovery process starts with computational screening of our large database of natural proteins.We combine cutting-edge deep learning models with traditional bioinformatics to predict and select candidate protein sequences suited for a given application.The process then mo

97、ves to the wetlab for high throughput strain engineering and functional validation of the selected proteins.High performing proteins are then advanced to larger-scale fermentations and pilot scale protein production to produce quantities sufficient for in-application testing.PRODUCT PIPELINE16TECHNO

98、LOGYDecoding Bio Snapshot2023Shiru has a B2B model with the aim to sell or license functional ingredients to ingredients,chemicals,or CPG companies.Shiru partners with select entities to help scale ingredient production and collaborateon the initial stages of ingredient development.Strategic partner

99、ships will enable Shiruto build relationships early with ideal future customers to ensure their needs are met.Shiru has two paths to market for the ingredients in their pipeline:1.Joint development with major global ingredients businesses to scale their products via licensing opportunities2.Direct s

100、ales of ingredients to companies developing consumer products across food and personal care.Shiru has more than a dozen active partners across these two GTM avenues.Shiru is also sampling its alternative fat solution with strategic partners.COMMERCIAL17Were Shiru.Were curious,ambitious,and a tad bit

101、 obsessed with our mission.Leadership:Jasmin Hume,PhDRanjani Varadan,PhD Jason Voogt,PhDMegan PittmanBoard of Directors Jasmin Hume,PhDDeena ShakirChuck TempletonMary ClarkeRachel KonradAdvisors Aaron KimballJim MillisDean BanksBlaine TemplemanBernhard van LengerichFounder&CEOCSO CPOVP of PeopleFoun

102、der&CEOLux CapitalS2G VenturesSVP at FirmenichCBO at The Production BoardFormer CTO at ZymergenFormer CTO at Natures FyndFormer CEO at TysonFormer CLO at Aduro BiotechFormer CSO at General MillsDecoding Bio Snapshot20232.2.INFRASTRUCTURE FOR BIO1819INFRASTRUCTURE FOR BIOWhy it matters:Whats changed:

103、How next gen infrastructure companies are different:If you look at recent examples of breakthroughs in AI/ML,the key bottleneck is always the data,rather than the models.For example,the Convolutional Neural Networks that drove the recent breakthroughs in computer vision were first studied in the ear

104、ly 90s.But the breakthroughs didnt happen until after the release of ImageNet in 2010.So the ability to collect and organize data will determine the winners and losers in the future of biotech.To do that well,you need infrastructure.-Biotechs are building data science teams that dont just require mo

105、re data,they require completely different approaches to collecting and storing data.Data stored in persistent,central location where it can be sliced and diced for re-analysis Contextual metadata,including experiment design and process must be captured in detail for communication to data scientists

106、and for later reference Formats must be consistent to allow analysis combining multiple experiments These types of analysis often require changes to experimental design outside what bench scientists are used to thinking about-An explosion of new instruments and types of readouts create abundant oppo

107、rtunities but with added complexity Every new type of instrument or readout must be integrated into the data infrastructure -LabOps and Automation provide opportunities to passively collect data that is much more consistent and reliable than otherwise.Experiment design data from automation logs is a

108、 much more reliable source of truth than anything bench scientists enter manually By harvesting metadata from these logs,bench scientists dont need to re-enter information-People from the Tech world are getting tired of click harvesting and moving into more impact-driven fields,particularly biotech

109、This will fuel the creation of separate data teams working in parallel to bench scientists Because they have less lab experience,collecting contextual metadata is more important to ensure clear understanding-They encourage bench scientists to collaborate with data scientists from the genesis of an e

110、xperiment,particularly on experimental design(Kaleidoscope,Benchling,Sapio,Scigilian)-They allow bench scientists to think about automation in terms of experimental design(Synthace,Automata,Strateos,Spaero)-They allow data scientists to quickly and reliably incorporate new data sources and integrate

111、 them with contextual metadata(Ganymede,Tetrascience)-They allow data scientists to seamlessly and repeatably run large-scale analysis on data from a wide range of sources(Latch,Watershed,Superbio,Pluto)Decoding Bio Snapshot2023AUTOMATATo transform the way scientists work&innovate by driving the ado

112、ption of automation at scale.Founding year:2016Founders:Mostafa ElSayedNumber of employees:155Location:London,UKLast funding round:Series BTotal funding raised:42MWere creating a fully-automated lab bench enabling labs to achieve higher throughput,total walkaway time and reliability,without compromi

113、sing on valuable lab space,while the accompanying lab orchestration software enables true walkaway time.Currently,lab automation tends to be defined by a series of very specialized benchtop instruments that are still reliant on human interaction.But Automata has a new way of looking at lab automatio

114、n in the:open,integrated automation.Integrated automation is when multiple devices are connected,both robotically and digitally,to turn workflows into one seamless automated process.This gives lab workers true hands-off time to focus on the work that really matters.Our automation is also open,which

115、means we are vendor agnostic:our system will integrate with the instruments labs already have or we will source the instruments best-suited to their workflows.Our mission is to work with every lab.However,were currently targeting two application areas,Genomics and Cell Culture,where weve seen organi

116、c traction and common workflows where automation adds substantial value to the lab.Core Technology:our offering consists of three components-Robotics Layer-the automation-enabled lab bench enables the reliable,automated transfer of lab-ware between any two instruments in a smart workcell,through a f

117、ully automated lab bench system-Digital Layer-a cloud-based laboratory orchestrator that seamlessly connects workflow activities-digital or physical;automated or manual-in one intuitive and customizable platform.-Services Layer-a dedicated Automation Scientist and Project Manager works with our cust

118、omers,from design to deployment,to align our automation solution to their goals.Current automation solutions come with significant compromises that prevent labs from scaling and adapting their workflows-ultimately preventing the labs long-term growth.Solutions on the market are often inflexible,diff

119、icult to scale and hard to use,making it a challenge for labs to truly access the benefits of automation today,and evolve their labs with it in the future.Automata solves this by designing with humans in mind,the modular bench fits into the same footprint as your regular lab benches.Both the robotic

120、s and software layers can be adapted for multiple workflows and are interoperable with multiple instruments,all while allowing for easy scalability within your existing environment.PRODUCT PIPELINE20TECHNOLOGYDecoding Bio Snapshot2023Market:Our mission is to work with every lab.However,were currentl

121、y targeting two application areas;Genomics and Cell Culture.Organically weve seen traction in these two areas,and both have common workflows where automation can add substantial value.Initial market sizing suggests the automation opportunity in both markets is in the$bns and growing double-digit per

122、centages year-on-year.Were focusing on Europe(primarily DACH and Benelux),the UK and the East Coast of the US,as our teams are based in UK and the East Coast of the US.Over 60%of the opportunity identified globally is in the US,so this is where we see the business in the long-term.10-Year Vision:Aut

123、omated labs are becoming experimental and diagnostic factories,generating outcomes at a scale,speed andquality orders of magnitude greater than what manual labs leverage today,Automata will be at the operational core of these labs driving both efficiencies in the lab and information upstream to the

124、tooling generating insights and improving the way science is done.Business Model:We develop and manage our platform in house(including the robotics,digital and services layers),however we deploy our platform to customers with 3rd party instruments(e.g.liquid handlers,centrifuges,etc).Were working on

125、 go-to-market partnerships with a range of instrument providers,but none are currently in place,so currently we sell directly to Life Sciences businesses,including Pharma,Biotech,CDMOs/CROs,and core testing labs.We offer various commercial models;including a standard upfront cost+service fees model,

126、as well as a usage-based model that incentivizes higher volumes.A key component of our business model is expanding alongside our customers;adding benches and instruments as their needs grow.Were targeting 20%repeat business from existing customers in 2023 and this will increase to 40%in 2024.Automat

127、a has been growing on average 170%YoY since 2021COMMERCIAL21Mostafa ElSayedSven PppelmannPantea RazzaghiKarian Lewis-KingPete SheaNick PattinsonCEOVP of EngineeringHead of DesignHead of PeopleCFO&COOVP of Product&StrategyManagement TeamDecoding Bio Snapshot2023CROMATICBuild softwares that will break

128、 down all technical barriers related to scientific outsourcingFounding year:2022Founders:Ann LinNumber of employees:8Location:San Francisco,CA,USALast funding round:Pre-seedTotal funding raised:$3MWere building an integrated platform for outsourcing research and development.Our market includes anyon

129、e in biotech looking to outsource work to a contract research organization(CRO).Our platform is applicable to biotechs of any stage-from early startups to established pharmaceutical companies.We take biotech clients all the way from initial proposal submission to project completion.Our platform will

130、 receive project proposals and quickly match biotechs to CRO candidates.Once a match is made,our platform facilitates regular communication between biotechs and CROs,allowing biotechs to more effectively track project progress.CROs present an opportunity for biotechs to increase the pace and scope o

131、f their work.However,coordinating experiments across entities can be challenging.Additionally,finding the right CRO for a request can be difficult.Biotech founders often rely on word-of-mouth when finding an outsourcing partner.However,this approach does not scale,and limits those new to the biotech

132、 space to the expertise of their immediate network.Our core technology is the Cromatic web platform which includes a marketplace to help biotech find CROs and software to help biotech collaborate with their external vendors on their R&D projects.We have a two-pronged approach for accelerating effect

133、ive outsourcing.First,the Cromatic marketplace will lower the barrier to identifying qualified CROs that have the bandwidth to take on new projects.Secondly,once a biotech finds a CRO match,our platform will facilitate regulator communication and milestone updates.Why is this a breakthrough?Currentl

134、y,no platform exists that combines effective CRO identification and subsequent project management.We give time back to biotech founders so that they can focus on their science.Our platform streamlines the otherwise difficult task of finding the right CRO for a biotechs project.More efficient outsour

135、cing may result in more projects being outsourced,allowing biotechs to run projects at an accelerated pace.PRODUCT PIPELINE22TECHNOLOGYDecoding Bio Snapshot2023Market:Outsourcing to CROs is currently estimated at over$100 billion in market value.With the growing wave of biotech startups,the outsourc

136、ing market is projected to triple within the next decade.Vision:The last few decades have seen many successful tech companies develop from nothing more than an individual coding at their computer.However,the need for physical lab space and specialized equipment has prevented a similar ease of growth

137、 in biotech.Cromatic seeks to give the same freedoms to biotech founders.We strive for a world where someone could launch a biotech company from their computer,using effective CRO management to complete research studies.Business model:We currently have two revenue streams.Biotech clients pay a upfro

138、nt annual subscription fee to assess our R&D marketplace and project management software.Vendors(CROs)pay a referral fee,a percentage on the total cost of the scientific request,on each successful contract they signed on the Cromatic platform.Partnerships:We have partnerships with LabDAO and Molecul

139、e,decentralized science collectives that funds early stage investigations.We also have a partnership with Nucleate,providing their teams with access to the Cromatic platform.We are about to finalize a number of other partnerships with incubators,accelerators,and venture capital firms,in which Cromat

140、ics services will be made available to their startups.COMMERCIAL23Decoding Bio Snapshot2023Ann LinAnne ChenPrasil LakshmananDrake GarnerKatherine LiCEOCTOSenior UX/UI DesignerScientific Partnership LeadLead EngineerThe TeamGANYMEDEBe the full-stack software backbone of biotechs and manufacturingFoun

141、ding year:2022Founders:Nathan Clark,Benson LeeNumber of employees:16Location:Fully Remote,United StatesLast funding round:Series ATotal funding raised:$15.5MGanymede cloud data platform data lake that connects instrument files and schematized tabular data,totally covering all data in your business a

142、nd relating it through the analysis.Includes dashboarding and automated reportingLab-as-Code integration pipelines everything in Ganymede is a dataframe,including the integrations.Write your own python code(or use ours out of the box as no-code)to move real data around from e.g.a plate reader to Ben

143、chling.Create no-code forms for scientistsAgents a variety of technologies for connecting to apps,lab instruments,etc.A fully browser based file sync tool,a Windows program,a physical IoT device we have,an API agent,an AWS/S3 IAM agent,etc.Coming soon:virtualization.Put entire apps into the Lab-as-C

144、ode pipelines to allow for human-in-the-loop interaction,such as doing FlowJo gating as an automated step.Coming soon:full self-service sign-up,create your own tenant at ganymede.bioLab-as-Code is revolutionary by merging the power of bioinformatics tooling with fully hosted infrastructure and perma

145、nent database creation as it goes,making Ganymede the computing model for integrating the wet lab.Uniquely,its also fully hosted no need to set up any infra.Ganymedes Lab-as-Code technology allows you to write code directly against your lab,and build no-code tools on top.Power tools for data scienti

146、sts:Lab-as-Code allows computational biologists to control the physical lab like a bioinformatics pipelineScalpels for lab scientists:automatically generate UIs for scientists and RAs/technicians to interact with automationBedrock of your data:Ganymede saves all data forever and versions it to make

147、compliance“just work.”Ganymede builds a data lake as you go and can be your complete software backendPRODUCT PIPELINE24TECHNOLOGYDecoding Bio Snapshot2023Nathan ClarkJohn LaSami BelkadiAlan ChramiecBenson LeeAndy CarusoFounderProduct,sales,engUX engineeringBackend eng,systemsProduct managementLab au

148、to,hardwareHead of scienceSales ops,recruiting,businessFounderScience,eng,hardwareHead of engineeringBackend eng,infraTargeting both mid-size biotechs(preclinical/process development,50-300 people)and large pharma manufacturing.Key criteria is that some physical/wet lab process is under developmentW

149、e exist to handle complexity.So we love science generally,e.g.small molecule,and biologics/synbio is even better.Custom instruments are greatWe have over$1M ARR committed and are targeting$4M in 2023We dont have any direct competitors-we see ourselves competing with AWS or GCP.Some companies offer i

150、nstrument integrations,but we have productized a platform to write and run integrations,the first of its kind;not to mention scientific analysis and database creation in the middleCOMMERCIAL25Leadership teamDecoding Bio Snapshot2023KALEIDOSCOPEKaleidoscope is building software to enable R&D teams to

151、 make their science more collaborative,reproducible,and scalable.Founding year:2021Founders:Bogdan Knezevic(CEO),Ahmed Elnaiem(CTO),David Yen(CPO)Number of employees:9Location:New York City,NY,USALast funding round:SeedTotal funding raised:$6MKaleidoscope is building software to make R&D more collab

152、orative,reproducible,and scalable.The Kaleidoscope platform allows scientific teams to easily organize and track scientific projects,key decisions,and experimental data.It sits on top of other software tools and provides intuitive but powerful ways to plan and track science(pound or sample-centric),

153、regardless of who was involved or where the underlying data is.Kaleidoscope also enables orgs to automate hand-off between people or teams,quickly move from ad-hoc to templated work,and easily manage permissions to share projects with external groups,collaborators,or partners.Today,most scientific t

154、eams do a very poor job of tracking work and codifying scientific decision-making.Biotechs most often either stitch together a hodgepodge of generic consumer or enterprise software,or they hire engineers to build custom tools in-house.The former simply isnt optimized for R&D work(underlying architec

155、ture is not built with a science use-case in mind,integrations with other vertical-specific tools dont exist,etc.)and/or breaks at the scale at which scientific projects are run.The latter is extremely expensive to build and maintain(in terms of both time and money),and most often results in a poor

156、product experience for the end user.By using Kaleidoscope,scientific teams can more easily track and version their projects from end-to-end(e.g.for replicating results or regulatory filings),automate cross-individual or cross-team workflows,and ultimately spend significantly more time and money focu

157、sing on the science at handPRODUCT PIPELINEEntity dataOperations planningAutomated triggersTrack data for any bio entity,across outside sources,in one placeSchedule assays and manage progression through a programEvent based automations to save teams time(coming soon)26TECHNOLOGYDecoding Bio Snapshot

158、2023We are a team of scientists,engineers,and technical product designers,with collective experience spanning everything from being early employees at acquiredstart-ups,to working at established tech institutions.Our beachhead market is biotech/pharma,with an initial therapeutics focus.In addition t

159、o this being a space that we as a team have experience working in,its also a market that is both maturing in its needs for more sophisticated tooling(due to increased decentralization of work,volumes of data being produced,etc.),and that has historically been underserved when it comes to specialized

160、 software.In 10 years,we want Kaleidoscope to become the go-to platform for collaborating,organizing,and tracking scientific decision-making and work.The problem abstractions were building around are consistent across a large variety of R&D-heavy fields,so we want to become the software that every R

161、&D company chooses for managing their internal workflows and scientific projects.Our mission is to help enable a world where all of scientific work is reproducible,no matter how many people or how much data was involved,and all of scientists time is spent meaningfully(ie focused on the actual scienc

162、e).As a SaaS company,we charge monthly/annual fees for customers to use Kaleidoscope for their work.We also occasionally run separately priced pilots for customers who need product customizations and who want to work with us over a fixed period of time to develop and test those features.This ensures

163、 that the product works as expected,for that customers use-case,before the customer commits to purchasing a software license.Currently,we are working with a spread of partners,including individual beta users,informal design partners,and paying pilot customers.COMMERCIAL27Bogdan KnezevicAhmed Elnaiem

164、David YenMandana ManzariAdam MephamStephanie ChengMaja VujicBen KieferTiffany Williams CEOCTOCPOBDEngineerEngineerEngineerProductOperationsKaleidoscope Team.Decoding Bio Snapshot2023SPAERO BIOSpaero is empowering scientists to program any liquid handler in their lab with confidence and ease.Founding

165、 year:2021Founders:Mitch AdlerNumber of employees:9Location:San Francisco,CA&Somerville,MALast funding round:SeedTotal funding raised:$5.2MSpaeros system is built on four technologies:Our intuitive UX gives scientists one simple tool that translates their experimental intent to any liquid handler in

166、 their lab.This decreases training time,increases usage time,and empowers scientists to automate more of their daily work.Spaeros compiler leverages sophisticated mathematical infrastructure to abstract away the complexity of programming a robot.This gives scientists the flexibility to automate expe

167、riments that require rapidly changing variables.A robot-agnostic design allows Spaeros system to control any liquid handler on the market.This dramatically reduces the time to adopt new robots and adds adaptability to a labs automation fleet.LiquidQC uses computer vision and AI to teach itself how t

168、o precisely and accurately dispense challenging fluids.This eliminates hours to months of tedious manual labor that would otherwise be required.PRODUCT PIPELINE28TECHNOLOGYScience today moves at the speed of human labor.The development of new medicines,new diagnostics,and new materials is currently

169、rate limited by manual processes.Labs have robots,but rarely use them.This is because the costs of automation outweigh the benefits.For instance,getting a robot programmed right takes weeks or months for a common experiment.Conducting the same experiment by hand takes hours.Spaero enables science to

170、 move at the speed of human thought.Our system enables scientists to tap into lab automation with confidence and ease.Flagship software:Experiment Builder allows scientists to flexibly describe and refine their experimental intent.Spaeros system then translates and runs that experiment on any liquid

171、 handling robot in their lab.Coming soon:LiquidQC,a modular hardware extension,allows scientists to automate workflows with challenging liquids.Decoding Bio Snapshot2023Mitch AdlerGabe JohnsonRobin WhitmoreAlex KreisherJay JackmanAmr AlyNitya NoronhaAlena LeeDimitris PapageorgiouCEOSoftware LeadProd

172、uct ManagerSoftware EngSoftware EngSoftware EngUX LeadLab Automation EngScience LeadMarkets:We are initially focused on R&D labs in Pharma and Biotech that need to scale their experimental throughput.The current state is heavily reliant on human labor.Automation is available-many labs have$100k+of l

173、iquid handling robots.However,these robots sit idle because these machines are often so painful to program making it faster to perform an experiment by hand.We see our system generalizing to a broad set of applicationsions including:drug discovery,chemicals,materials,foods,and other products with th

174、e potential to transform our daily lives.Vision:The pace of scientific discovery has been dramatically accelerated by the increase of robot-facilitated R&D.Through Spaero,robots have become the default for performing repetitive tasks,and scientists are free to focus on innovative work.Spaero is the

175、ubiquitous tool that enables humans to translate their experimental intent to any robot in their lab.Business model:We have a software as a service model with pricing tiers based on usage.This aligns our revenue directly with our customers need to scale their experimental throughput.COMMERCIAL29The

176、TeamDecoding Bio Snapshot20232.3.AI-DRIVEN DESIGN30Why it matters:Whats changed:The ability to engineer protein therapeutics with novel structure and function has been a long-standing goal in drug discovery.In the past year,academic labs and biotech companies have leveraged methodological advances i

177、n a subclass of generative algorithms called diffusion models to vastly improve de novo(i.e.,from scratch)protein design.Diffusion models enable a big leap forward in programmable protein engineering,where proteins with a wide-array of user-specified structures and functions can be rationally genera

178、ted.In the non-bio machine learning world,diffusion models were first popularized for image synthesis tasks in a 2020 paper from Pieter Abbeels group,and quickly surpassed generative adversarial networks(GANs)as the workhorse generative model.Briefly,diffusion models function by iteratively adding g

179、aussian noise to training data,and then learning to recover(or generate)the data by reversing this process.After the diffusion model is trained,new data is generated by passing randomly sampled noise through the trained model.Importantly,these models can be guided throughout the generative process t

180、owards specific design objectives(such as a desired structure or function)by providing conditioning information to the model.Check out this explainer of diffusion models for more detail.31AI-DRIVEN DESIGNDecoding Bio Snapshot2023Figure:Examples of protein structures designed from scratch using diffu

181、sion models.From Ingraham et al.,2022 published as a preprint by Generate Biomedicines.32Decoding Bio Snapshot2023Whats different,and where the field isat today:There has been a flurry of papers over the last year applying diffusion models to protein engineering.To highlight a few:-The first preprin

182、t published(all the way back in May 2022,that is how fast this field is moving)described a protein structure and sequence diffusion model.The model was tested on a range of tasks such as structure inpainting(generate a full protein given a small structural motif)and immunoglobulin loop backbone and

183、sequence generation.-Generate Biomedicines combined diffusion models and graph neural networks to generate full multi-protein complexes,and can be conditioned to design proteins with specific properties and functions(see figure).-The Baker Lab developed a protein engineering system that combines str

184、ucture prediction networks and diffusion models.By fine tuning the RoseTTAFold protein folding algorithm on a protein structure denoising task(creating a new model called RFdiffusion),strong performance is obtained on a range of tasks including protein binder design and enzyme active site scaffoldin

185、g.The experimental success rate(i.e.,the proportion of experimentally synthesized and tested proteins that confirm the in silico prediction)of RFdiffusion was improved by two orders of magnitude relative to previous algorithms developed by the Baker Lab.Similar to what we saw after the release of Al

186、phaFold,it will be exciting to see how far the field can push the diffusion models in protein design.For example,can these systems incorporate nucleic acids to generate DNA or RNA binding proteins?Or incorporate ligands to optimally design small molecule binders for target proteins?Not to be left be

187、hind by their biologic counterparts,there has also been exciting progres in generative AI for small molecule design.PostEra has developed an end-to-end machine learning platform for medicinal chemistry.One of PostEras most innovative methods is automated retrosynthesis.A transformer algorithm is use

188、d to predict the outcome of reactions.This granular understanding of synthesis is combined with large proprietary datasets and novel approaches to molecular design and active learning to create a closed-loop Design-Make-Test cycle.Demonstrating the power of the approach,the platform has been used to

189、 design a SARS-CoV-2 protease inhibitor.ABSCIAbsci is a generative AI drug creation company with a mission to create better biologics,faster.Founding year:2011Founders:Sean McClainNumber of employees:200Location:Vancouver WA,New York,NY,and Zug SwitzerlandLast funding round:IPO Jul 2021Total funding

190、 raised:$425MWere building an Integrated Drug Creation Platform for developing better biologics faster,our partnered assets are rapidly moving into clinic.Our market includes pharmaceutical,biotechnology and technology companies looking to accelerate and innovate on biologics drug discovery.Ultimate

191、ly the market includes payors and patients who will benefit from the better biologics created through our Integrated Drug creation platform.We can accelerate time to clinic and increase probability of success for biologics drug discovery using generative AI.Our approach is data to learn,AI to create

192、 and wet lab to validate.Our scalable proprietary wet-lab technology generates billions of data points which is used in training our generative AI engine.This addresses the problem of limited scalable biological data which has hindered generative AI for protein drug discovery.Our generative AI engin

193、e can create optimized designs for therapeutic compounds and our wet-lab infrastructure can validate hundreds of thousands of these AI generated designs within weeks.Weve successfully operationalized this AI and wet-lab integration for rapidly iterating and improving our models.PRODUCT PIPELINE33Dec

194、oding Bio Snapshot2023Traditional drug discovery takes on average 10 years and$1.3B to get a drug from research to approval with a success rate of under 5%.Leveraging Abscis platform we can accelerate time to clinic,increase probability of success through multiparametric optimization of biologics an

195、d reduced time to clinic also decreases cost of drug development.Our Integrated Drug Creation Platform leverages generative AI and synthetic biology to create better biologics faster(inclusive of de novo discovery in silico,AI lead optimization in silico,novel antibody and target discovery).Our appr

196、oach is data to train,AI to create,and wet-lab to validate.Why is this a breakthrough?Generative AI is being leveraged in many other sectors but has previously had limited success in protein drug discovery because of the lack of scalable biological data.Absci has solved the problem of scalable biolo

197、gical data with proprietary wet lab technologies that can screen billions of protein-protein interactions per week,feeding our generative AI model,and subsequently validating AI designs in our wet lab.This iterative cycle is what allows us to accelerate time to clinic for drug development and increa

198、se probability of success.Absci generates and harnesses data at scale.Our proprietary wet lab ACE assay allows us to screen billions of cells per week,which generates high quality data that trains our generative AI engine that creates novel drug designs which are subsequently validated in our wet-la

199、b.This cycle is done within weeks and enables biological data at scale.34TECHNOLOGYDecoding Bio Snapshot2023Market:Our partners are large pharmaceuticals,biotechnology companies,and technology companies looking to accelerate and innovate on biologics drug discovery.Vision:Our vision is fully in-sili

200、co drug creation where we can create breakthrough therapeutics at a click of a button,for every patient.Business model:Our drug discovery and development partnerships have upfront payment,milestone payment as the drug candidate moves through the clinic,as well as commercial milestones and royalties.

201、Our model aligns us with partner incentives with upfronts as well as partaking in the long-term success of drug candidates discovered and developed with Abscis technology.Absci is also pursuing developing an internal pipeline of assets to IND and clinical proof of concept which validates our platfor

202、m as well as providing better biological drug candidates with efficacy signals with value in the asset themselves.Weve have several partnerships underway including a$610M+royalties collaboration with Merck for drug discovery and a multi-target drug discovery deal with EQRx among others.We also have

203、a strategic partnership with Nvidia to accelerate in silico drug creation.COMMERCIAL35Decoding Bio Snapshot2023Sean McClainFrans Van HoutenZach Jonasson,PhDJoseph Sirosh,PhDDan RabinovitsjKaren Mcginnis,CPAAmrit NagpalKarin WierinckJack GoldPenelopeAndreas Busch,PhDGreg Schiffman,CPASarah Korman,PhD

204、,JDFounder,CEO&DirectorFormer CEO,PhiiipsManaging Partner,PVPVice President,Alexa Shopping,AmazonVice President,Connectivity,Meta(formerly Facebook)Former CAO,IlluminaManaging Director,Redmile GroupChief People OfficerChief Marketing OfficerChief Morale OfficerChief Innovation OfficerChief Financial

205、 OfficerChief Legal OfficerExecutive Leadership TeamBoard Of DirectorsCRADLEDemocratize protein engineering using Generative AIFounding year:Q421Founders:Stef van Grieken,Jelle Prins,Elise de Reus,Eli Bixby,Harmen van RossumNumber of employees:12Location:Zurich,Switzerland&Delft,NetherlandsLast fund

206、ing round:SeedTotal funding raised:5.5MCradle uses generative AI models to engineer proteins that we make available to anyone through easy to use web-application,apis and software libraries.Why is this a breakthrough?Existing evolution,homology and energy based methods are error prone and hard to us

207、e.Generative models significantly outperform these methods on common protein engineering tasks(function,expression,stability,solubility,binding affinity,activity etc).Though protein engineering is a critical component of developing any bio-based product less than 5%of designs tested in the lab achie

208、ve their design targets,while being one of the biggest cost drivers.We dramatically improve the success of producing a sequence that hits your design goals.Cradle reduces the time and cost to get to a protein with the desired properties.We dont keep that technology for ourselves,but make it availabl

209、e to anyone.Success would mean small teams can dramatically punch above their weight.For engineering stability of a protein we benchmarked our Generative AI models against energy based methods(Rosetta)on 87 empirical studies that are broadly distributed in protein space(enzymes,peptides,antibodies e

210、tc).We are also validating our in-silico generated sequences with 2 design partners and on an in-house project on T7 RNA polymerase with positive results.36TECHNOLOGYDecoding Bio Snapshot2023Market:we are piloting our product with two design partners(listed US pharma&scale up natural product synthes

211、is company).After coming out of stealth Nov 22 over 250 companies and research organizations signed up for early access.Our focus is on lead discovery&optimization for antibodies,but we believe the technology to be broadly applicable across protein engineering in pharma,chemicals,food,materials,etc.

212、Vision:We enable any small team to discover and engineer a protein with a few million dollars without a wet-lab across different industries.Think Figma for proteins.Business Model:we license the software as a service.We may also provide pay for experiment for common experiments(i.e.thermal shift ass

213、ay,solubility assay,binding affinity(ELISA)assay,monomer levels(SEC-HPLC)on demand.COMMERCIAL37Decoding Bio Snapshot2023Stef van GriekenJelle PrinsMax HenningssonSytske BesemerMartin TschechneArthur LindoulsiTomasz SodzawicznyEna kopeljaEmily SoleyElise de ReusFranzi GeigerAdam WeberDaniel DanciuHar

214、men van RossumEli BixbyCEODesignDesignDesignMachine LearningMachine LearningEngineeringMachine LearningBioengineeringBioengineeringMachine LearningEngineeringCTOBioengineeringMachine LearningThe TeamLABGENIUSConventional antibody discovery methods involve the sequential optimisation of different pro

215、perties through rational design.With this approach,improving one property can inadvertently worsen others.These trade-offs inevitably lead to costly failures or sub-optimal patient outcomes.At LabGenius,weve developed an iterative search process that combines high-throughput experimentation with ML-

216、led design to systematically explore and more efficiently discover novel high-performing therapeutic antibodies(co-optimised across multiple features e.g.potency,efficacy,selectivity,and developability)LabGenius is a next-generation antibody discovery company.Through the integration of high-throughp

217、ut experimentation and machine learning,the company has developed a new way of searching antibody design space for high-performing molecules.38TECHNOLOGYTo combine the best of human and machine intelligence to accelerate the discovery of uniquely powerful antibody therapeuticsFounders:James Field,Ph

218、DNumber of employees:50 Location:London,UKLast funding round:Series ATotal funding raised:$27.5MThe first molecules in our pipeline are T-cell engagers(TCEs)for the treatment of solid tumours.Challenge:The effective treatment of solid tumours is a significant unmet need.TCEs are a type of engineered

219、 antibody that redirect the immune systems T cells to recognise and kill cancer cells.Investment into the discovery of new TCEs has been significant and there are currently 60 different CD3-targeting TCEs for solid tumours being investigated in clinical trials.Unfortunately,the progression of TCEs t

220、hrough clinical trials has been plagued by issues with dose-limiting on-target,off-tumour toxicity.This occurs when healthy cells expressing a tumour associated antigen(TAA)get caught in the crossfire.Solution:Our ML-driven protein engineering platform can systematically explore design space and ide

221、ntify high-performing antibodies-often with non-intuitive designs.We are leveraging this capability to identify safer and more effective TCEs for solid tumours that do not suffer from issues with on-target,off-tumour toxicity.PRODUCT PIPELINEDecoding Bio Snapshot2023COMMERCIAL39Partnerships:In addit

222、ion to financing the development of an internal pipeline,LabGenius also selectively engages with leading pharmaceutical and biotech partners to pursue both technology and asset development opportunities.For example,LabGenius has an on-going multi-year collaboration with pharmaceutical company Sanofi

223、.In this collaboration,LabGenius is applying its ML-driven antibody engineering platform to the co-optimisation of therapeutic NANOBODY proteins.The LabGenius team is 50 people in size and includes experts in drug discovery,synthetic biology,computer science and automation.The company is led by an e

224、xperienced Management team and Board that benefits from the support of world-class investors and advisors.All information as of Q1 2023.James Field,Ph.D.James Field,Ph.D.Chris GibsonPatrick Pichette Tobias Arkenau,Ph.D.Victor Greiff,Ph.D.Aaron Kimball Gino Van Heeke,Ph.D.Leonard Wossnig,Ph.D.Edwin M

225、oses,Ph.D.Irina Haivas Shaquille Vayda Rohan GaneshFounder,Chief Executive OfficerFounder,Chief Executive OfficerRecursion Founder&CEOEx-Google CFO,Ex-Twitter Chairman CMO&Global Head of Drug DevelopmentAssoc.Prof.for Systems ImmunologyBenchling Head of Engineering Chief Scientific OfficerChief Tech

226、nology Officer Chairman Investor DirectorInvestor Director Investor Director BoardManagementAdvisorsWhats different about the way in which LabGenius searches antibody design space?In contrast to conventional approaches,we use an active learning method called Multi-Objective Bayesian Optimisation(MOB

227、O)to efficiently navigate design space and find high-performing antibodies.This approach is largely free from human bias and so the novel antibodies we discover frequently have highly non-intuitive designs.A key aspect of our approach is that our models are trained using data from disease-relevant f

228、unctional cell-based assays.Our focus on functional assays with high predictive validity gives us the best chance of finding molecules that will perform well in the clinic.How are LabGenius capabilities world-leading?Through the integration of high-throughput experimentation with predictive modellin

229、g,weve established a screening throughput that is world-leading.In just 12 weeks,we can experimentally screen 800 T-cell engagers in functional cell-based assays.By combining the resulting characterisation data with our ML models,we can reliably evaluate 28,000 variants virtually.With this capabilit

230、y we can rapidly and reliably deliver novel TCEs with best-in-class selective killing profiles when compared to the corresponding clinical benchmarks.In a recent demonstration project,we used ML to systematically deliver highly efficacious TCEs that demonstrate 400-fold tumour killing selectivity ve

231、rsus the clinical benchmark*The predictive validity of an assay is hard to quantify but it is commonly accepted that for T-cell engagers,killing assays in tumour derived cell lines have a much higher predictive validity compared to simple binding assays.Decoding Bio Snapshot2023POSTERAWe are buildin

232、g the worlds best Chemistry AI platform to bring more cures to patientsFounding year:2019Founders:Aaron Morris,Alpha Lee,Matthew RobinsonNumber of employees:27Location:Boston,MALast funding round:Series ATotal funding raised:$26MPostEra is running commercial programs in virology and oncology both vi

233、a an internal pipeline and also in partnership with partners like Pfizer and the NIH.Our most advanced programs are at Lead stage with the expectation to announce our next Preclinical candidate in the coming 6 months.Additionally,PostEra launched COVID Moonshot,which became the worlds largest open-s

234、cience initiative to develop a COVID antiviral in an open-science fashion.The project developed from a fragment screen to preclinical candidates which are now going through IND-enabling studies.We expect to launch a Phase I study in 2024 as the worlds first crowdsourced antiviral.The drug is intende

235、d to be low-cost and an important line of defense in developing nations where vaccine access is limited.PostEra is building Proton;an end-to-end medicinal chemistry platform powered by machine learning.The process of medicinal chemistry constitutes a 3-step Design-Make-Test cycle and the foundation

236、of PostEra is a series of academic breakthroughs that unifies a machine learning approach across all three stages.One of the most significant breakthroughs is in the area of automated retrosynthesis.Making molecules is a rate limiting step in the preclinical stage and PostEras Manifold platform,a su

237、bset of Proton,productizes a core ML model called Molecular Transformer(MT)to predict the outcome of reactions with a performance on par with human chemists meaning we can scale a synthesis-driven design approach across all our small molecule programs.We combine MT with the worlds largest CRO databa

238、ses for which we have a dynamic view of global available inventories and building blocks.We combine this very granular understanding of synthesis with large proprietary datasets and novel approaches to molecular design and active learning to create a closed-loop Design-Make-Test cycle.The upshot,is

239、that PostEra is able to take a 5-year preclinical process and condense this down to 18 months which has been proven in real-world drug discovery programs such as COVID Moonshot.PRODUCT PIPELINE40TECHNOLOGYDecoding Bio Snapshot2023PostEra aims to be the industrys leading 21st century biopharma,with a

240、n AI-first approach to drug discovery and a tightly-integrated culture between engineers and scientists.We aim to bring new cures to patients and inspire a new generation of computationally-driven biotechs.Our business model is intentionally evolutionary over time in terms of the ratio of internal v

241、s partnered drug discovery programs.That ratio is currently 50:50 and will likely tilt toward a higher ratio of internal programs over time.This allows us to build a diversified portfolio over novel biology targets with partners and more biologically validated targets internally.Partnerships:in 2021

242、 PostEra signed a$260M AI Lab partnership with Pfizer to advance multiple drug discovery programs with an initial focus in oncology and COVID-19 antiviral therapeutics.PostEra is also the recipient of a$68M grant from the NIH to co-lead an AI antiviral drug discovery centre for the prevention of fut

243、ure pandemics.COMMERCIAL41Decoding Bio Snapshot2023Aaron MorrisTony SchroederMarie Laure RivesMihir TrivediJames FrickAlpha LeeMatthew RobinsonSasha ValleEmily RipkaCEOHead of MLHead of BiologyProduct LeadML Technical LeadCSOCTOChief of StaffHead of ProductThe TeamREDESIGN SCIENCETo discover a drug

244、for every target by using simulation and AI to bridge the gap between basic research and applied researchFounding year:2017Founders:David Rooklin,Haotian LiNumber of employees:30Location:NYCLast funding round:SeedTotal funding raised:$17MTraditional in-silico methods fail when confronted with challe

245、nges like low-fidelity structure data,functional conformational changes that are important to druggability,and large-molecule interactions like PPIs and nucleic acid systems;however,these challenges apply to the vast majority of the proteome.Our platform relieves a critical bottleneck between basic

246、research and applied drug development,opening the entire proteome to rational discovery.Redesign Science is building the first generative AI for drug discovery to unblock,accelerate,and scale preclinical biotech.Our AI is trained on massive data generated by first-principles physics,giving us access

247、 to training data others cannot reproduce or purchase.Why is this a breakthrough?Pure AI approaches have too little high-quality structure data to train their models,and commercially-available pure physics approaches lack both scalability and the capacity for generation of enriched,intelligently-ann

248、otated atomic datasets.By creating a flywheel connecting proprietary physics methods,massive data generation,deep learning,and feedback from experiment,we unleash the true power of both AI and physics-based approaches to rational drug discovery in an escalating cycle of increasing breadth,accuracy,a

249、nd speed.PRODUCT PIPELINE42TECHNOLOGYWe develop small molecule drugs against emerging and challenging targets.Our programs range from wholly-owned to co-owned to partner-owned assets with economically meaningful up-front,milestone,and royalty agreements.Our current pipeline is focused on oncology an

250、d autoimmune/inflammatory indications;however,we are not limited to these indication spaces.Why focus here?oncology and inflammation/immunology indications present large unmet medical need and a space of well-validated targets to work from,while offering commercial opportunity in combo therapy,rare

251、disease,and strategic partnership with big pharma and mid/large-cap biotech.Decoding Bio Snapshot2023Market:we find small molecule drugs for challenging and emerging targets associated with high unmet medical need,with a goal of bringing first-in-class oncology and inflammation/immunology drugs to t

252、he clinic.We have a unique advantage in small molecule discovery and are poised to exploit targets with well-validated biology and therapeutic relevance but for which no drugs currently exist due to druggability challenges.Vision:Our vision is a world where new breakthroughs in basic biology researc

253、h(new understanding of molecular pathways,metabolic dysfunctions,transcription processes,etc)lead immediately to drugs that take advantage of that knowledge to save lives.Business Model:We are advancing 8 internal,wholly-owned projects and 2 co-development partnerships,with plans to scale to additio

254、nal partnerships.This combination model enables us to internally identify un-druggable or under-served targets and indications,while gaining exposure to novel or emerging un-druggable targets partners have identified.COMMERCIAL43Cofounder/CEOCofounder/CTOGeneral councelPrincipal methods scientistVP

255、of MLInfrastructure leadVP of DataOpsMethodology ScientistsDirector of MLML scientistsEngineers,Infrastructure devsDirector of BDVP StrategyDirector BiologyDirector Med-ChemVP Molecular modelingApplications scientistsThe TeamThere are 20,000 genes that code for proteins in the human body,but fewer t

256、han 700 of them are targeted by available drugs.Our platform unlocks the druggability of the human genome.Results of experimental validation in our first four programs:-We discovered and optimized hit compounds for a formerly undruggable protein-protein interaction with favorable potency and minimal

257、 adverse effects in mouse models.-We identified mutant-selective hits to recover the protein-DNA interaction of a well-validated transcription factor implicated in multiple cancers.-We discovered PPI-inhibiting hits on a signaling factor implicated in many cancers by targeting a novel cryptic pocket

258、 we discovered-We discovered hits that allosterically modulate a well-validated apoptotic factor implicated in multiple hematological cancers,via a novel mechanismDecoding Bio Snapshot2023TERRAY THERAPEUTICSTerray was founded with the thesis that the biggest bottleneck in AI-driven drug discovery is

259、 insufficient access to high-quality chemical data.Generative AI algorithms are only as powerful as the data on which they are trained.To solve this,we built a platform that combines synthetic and medicinal chemistry,biology,automation,nanotechnology and AI to generate the largest dataset in existen

260、ce that maps quantitative interactions between small molecules and drivers of disease(targets)such as proteins or RNA.The platform uniquely combines computation and ultra high-throughput experimentation:Experimentation:-We have developed a method to rapidly and efficiently produce the worlds most ul

261、tra-dense small molecule microarrays displaying a combinatorial library of millions of molecules in a chip smaller than a nickel.The quantitative binding affinity for the entire library can be measured against a target of interest in just minutes using a fluorescent readout.Using our pre-built diver

262、sity compound collection,we measured over 1.5 billion unique small molecule/target interactions last year,backed by 40 billion individual measurements and are accelerating,making our data set uniquely powerful for data-hungry computational approaches like generative molecular design.-We pair this wi

263、th the ability to build computationally designed follow-up libraries in a matter of weeks to quickly pursue lead optimization.From there,computational algorithms are used to determine which molecules should be rapidly tested for activity in biochemical and cell assays.The automated resynthesis platf

264、orm is able to synthesize hundreds of prioritized compounds in just a few days for high throughput biological testing such as cell activity and in-vitro ADME.-In addition to the refined molecules and computational models developed in the iterative cycles of our platform we leverage additional comput

265、ational tools such as structure based design to power final optimization of molecules to promising leads for clinical translation.44TECHNOLOGYFounding year:2018Founders:Jacob Berlin,PhD,Eli Berlin and Kathleen Elison,PhDNumber of employees:75 Location:Monrovia,CALast funding round:Series ATotal fund

266、ing raised:$81.4MAt Terray,were improving human health by transforming the speed,cost and success rate for small molecule drug development using novel experimental data streams at scale paired with computation.Weve built an ultra-high throughput experimentation platform that generates chemical data

267、purpose-built for generative AI.With much faster iteration times,orders of magnitude larger scale and better precision than any other chemical data out there,we power modern computation to derive insights that rapidly lead to new drugs for patients in need.Our internal pipeline is focused on Immunol

268、ogy and we work in partnership(with upfront payment,milestones,and royalties)with large pharma and biotech.Were bringing drug discovery into the Information Age.PRODUCT PIPELINEDecoding Bio Snapshot2023The Drug Discovery Engine Fueled by Premium Chemical DataComputation:-The foundation of Terrays co

269、mputational capabilities are scale+quality+iteration powering advanced deep learning models for ligand-and structure-based drug design:Scale:billions of unique target-ligand interaction data points allow us to continuously expand our map of the molecule and protein universe Quality:each data point i

270、s supported by 20-35 repeat measurements,resulting in precise,IC50-quality data that unlocks accurate molecular property prediction with deep learning regression models Iteration:a rapid and efficient design-make-test-analyze cycle enables an active learning approach that intelligently and smoothly

271、explores chemical spaces millions of molecules at a time-The combination of these fundamental pillars allows us to unleash the power of advanced deep learning models for predictive,iterative molecular design-In our quest to efficiently and thoroughly explore chemical space,weve developed our own smo

272、oth,continuous,differentiable,and invertible representation of chemical space This embedding is inspired by state-of-the-art methods for LLMs,which have been used very successfully to represent both natural language and biochemical data,but incorporates insights from computational chemistry to accur

273、ately represent drug-like molecules-This allows us to perform multi-parameter molecular optimization on the surface of differentiable,constrained regressors trained on billions of datapoints,turning molecular design into a mathematical optimization problemOur experimental speed,scale and precision a

274、llow for differentiated generative AI capabilities which gives us a unique opportunity to identify and develop better compounds with a higher likelihood of clinical success to deliver novel medicines to patients in need.45Decoding Bio Snapshot2023COMMERCIAL46Our Challenge:Drug discovery needs more c

275、hemistry.Chemistry is the key to drug discovery,but chemical data is stuck in the twentieth century.Most current drug discovery platforms are built on data that is subscale,poor quality or both.And even the most powerful generative AI cant get past insufficient or incomplete data.Its time to address

276、 this problem at the root cause.Find the signal in all the noise.And generate data that bring quantity and quality together so we can build stronger models and predict better molecules.Our Innovation:Premium chemical data at unrivaled scale.Were generating chemical data purpose-built for drug discov

277、eryand were doing it on a larger scale and faster than has ever been possible.With orders of magnitude better resolution than any other data out there,we can answer a vast,unprecedented array of questions,meeting them with the power of modern computation to derive insights that rapidly lead to new d

278、rugs for patients in need.Were bringing drug discovery into the Information Age.Our Platform:The chemistry engine to solve hard problems.Our engine works better because it runs with scale,precision and iterative capabilities no one else can reach.We run iterative cycles of virtual molecular design a

279、nd precise biochemical measurement,feeding forward every experiment to power artificial intelligence and machine learning models,which in turn guide the next cycle of design.Our system runs with unmatched speed and scale,becoming more precise with each cycle.This is how we accurately navigate the va

280、st chemical space to deliver therapeutics for previously inaccessible targets.Our Vision:Transformative therapeutics at our fingertips.Were building a future where the speed,precision and scale of our engine enables us to predictably create drugs that meet urgent patient needs.Our Business Model:We

281、are a drug discovery and development company advancing our internal pipeline and also working in partnership with large pharma and biotech.For our internal programs we are focused on Immunology.We target areas where selectivity has been the core limitation,where our platform is uniquely advantaged.I

282、n the next few years,we are aiming to bring multiple candidates into the clinic with many pre-clinical opportunities behind them provided by our proprietary integrated experimental and computational platform.In addition to our own programs,partnerships in alternative areas of biology and unmet medic

283、al need are an integral part of our business model.In October 2022,Terray entered into a large,multi-target collaboration with Calico.Terray and Calico will identify small molecule leads against a set of targets nominated by Calico using the Terray tNova platform,with Calico subsequently assuming re

284、sponsibility for development and commercialization.Terray received an upfront payment and is eligible to receive milestones as well as tiered royalties on net sales.Jacob Berlin,PhDJacob Berlin,PhDBrian Kotzin,M.D.Ian Watson,PhDJohn Maraganore,PhDChristine Humblet,PhDDon Payan,M.D.Fiona Black,PhDChr

285、is PicardoGeoffrey SmithDusan PerovicEli BerlinBassil Dahiyat,PhDSudha Parasuraman,M.D.Fez Ujjainwalla,PhDEli BerlinCraig Schulz,PhDCarl Ebeling,PhDNarbe Mardirossian,PhDVanessa Taylor,PhDCo-founder and CEOCo-founder and CEOMadrona VenturesDigitalis VenturesTwo Sigma VenturesCo-founder,CFOand COOInd

286、ependent Director,CEO of XencoreIndependent Director,CMO of Ribon TherapeuticsHead of BusinessCo-founder,CFOand COOHead of AutomationHead of ProductionHead of Computational and Data SciencesHead of Biology and Preclinical DevelopmentExecutive Leadership TeamBoard of DirectorsSAB&Strategic AdvisorsDe

287、coding Bio Snapshot20232.4.TARGET DISCOVERY&VALIDATION47TARGET DISCOVERY&VALIDATION48Decoding Bio Snapshot2023Why it matters:Whats changed:Target discovery involves deciding on which proteins or genes,when inhibited or agonized,are most likely to induce a desired clinical effect on a patient populat

288、ion.This is an critical area of drug development,as almost everything else is downstream of this,and doing this in a differentiated way can majorly improve chances of success of a drug.We are highlighting this area as it is a cornerstone of drug development,and one that is fast evolving to allow bet

289、ter risk assessment of targets pursued by biotechs.The rise of next-gen target discovery methodologies and companies stems from several factors,including the effectiveness of reverse translational approaches,the rise of affordable and scalable single-cell genomics technologies,new machine learning m

290、ethods that allow for a wide set of predictions that can be used to evaluate targets,and more.How new target discovery&validation companies differentWe have observed that next generation biotech companies approach target discovery and combination therapy prioritization in ways that are distinctive f

291、rom traditional biotechs.Some of these approaches are also attempted by some larger pharma and traditional biotechs,and we are highlighting here some relative differences.Some of the ways:-Next gen biotechs leverage a wide variety of data sources,such as data from own trials,data from papers and var

292、ious omic modalities.Many larger pharma companies are also increasingly moving in this direction,and in fact have an advantage of their own large datasets.However,one particularly differentiating factor is that many next-gen biotech companies model these data jointly to understand the desired target

293、,phenotype,or drug combination(eg.methods in single-cell that jointly model different sets of omic modalities).The hypothesis here is that such joint modeling leads to a richer understanding of the phenotypes of interest,and eventually better results in tasks such as target prioritization.We can thi

294、nk of these as the first steps toward foundational models in biology.-Next gen biotech companies increasingly allow machine learning to play a driving role in designing experiments by leveraging in silico predictions(eg scGen)of experimental results.One can,for example,simulate in silico numerous ex

295、periments(of gene up/down regulation),and use the results to prioritize which experiments to run in vitro or in vivo.That is,the experiments for which the in silico results were most encouraging.This leads to fewer and more informative experiments,hence allowing for faster discovery and validation.T

296、his also allows for faster iterations of experimental loops,each time improving the MLs predictive accuracy.-Next gen biotech companies are organized and technologically prepared for quickly moving from in silico discovery to in vitro and animal testing.In many cases this takes fewer than two weeks.

297、They are organized as highly multidisciplinary teams that enable this quick iteration.-Next gen biotech companies leverage large scale reverse translational approaches.This constitutes using human data to generate new hypotheses to find new targets.New biotechs,such as Verge Genomics,lean on human d

298、ata more heavily than pharma/biotech do from the start of the discovery process.This is due to their advantage in analyzing large datasets,despite having smaller proprietary datasets.These approaches are believed to increase chances of success compared with relying relatively more heavily on animal

299、experiments(as is being increasingly endorsed by the FDA).ARPEGGIODrug oncogenic transcription factors.Founding year:2018Founders:Joey Azofeifa,Tim Read,Laura NorrisNumber of employees:15Location:Boulder,CO USALast funding round:Series ATotal funding raised:$20MWe develop small molecule inhibitors t

300、o transcription factors(TFs),a class of historically undruggable disease-associated proteins.Our platform can discover novel inhibitors of many different TFs that are associated with disease in a variety of indications such as autoimmunity,cancer,and neurodegeneration.Weve decided to focus on TFs th

301、at are associated with cancer,because the path to the clinic is a little more straightforward(toxicity is a smaller concern,relatively good preclinical models,large patient populations).Transcription factors(TFs)are proteins that regulate transcription of hundreds of genes and represent some of the

302、most well-validated targets for human disease.However,these proteins are notoriously challenging to drug by traditional small molecules(much of med chem focuses on kinases and other structurally-ordered proteins).Our technology can find novel actuators of these proteins potentially unlocking a whole

303、 new class of medicine and bringing drugs to diseases that havent seen improvements in decades.Transcription factor proteins are extremely hard to drug and our platform is finding novel chemistry to modulate them.Weve developed GRETATM that can rapidly screen large compound libraries in live cells f

304、or changes in the transcriptome.Our phenotypic screen identifies novel compounds that elicit specific changes in the transcriptome.From there,we perform rapid custom chemistry on these scaffolds to jointly optimize for transcriptomic effect and drug-like properties.We are assembling a large dataset

305、linking small molecular structure and transcriptomic activity and use machine learning algorithms to in silico design small molecules with specific transcriptomic properties.Why is this a breakthrough?Up until GRETATM,transcriptomics was too expensive to scale into a drug screen.With new improvement

306、s in sequencing,library prep techniques,and lab automation,weve developed an extremely cheap and high-throughput transcriptomics platform that will bring RNA-seq outside profiling 100 to 10,000 samples at a time.Biology is an extremely complex process.By leveraging high content phenotypic screening,

307、we can blackbox biology and just look for disease modifying chemistry.The difference for Arpeggio is that our“blackbox”measures over 20,000 pieces of RNA,so this allows us to capture the complexity of Biology without needing to understand how all the parts are linked together.PRODUCT PIPELINE49TECHN

308、OLOGYDecoding Bio Snapshot2023Historically,much of pharma has focused on drugging proteins that are highly ordered and structured.Traditional targets like kinases are very druggable both at the ATP-binding pocket and other allosteric sites.However,we are noticing a decline in the productivity of lar

309、ge Pharma,in part because many of these low hanging fruits of proteins classes have been picked.We believe that by developed technology that addresses a completely different protein class such as transcription factors we will be able to open a new market and build an IP moat around the way that we c

310、reate compounds to inhibit and actuate these targets.Ultimately this will usher in highly differentiated products for patients and address new disease markets such as untreated forms of cancer,autoimmunity,and neurodegeneration.Vision:Typically when we think about drug discovery,we think about desig

311、ning a small molecule to inhibit one protein.But what if you could design a compound to change the entire network?With enough transcriptomic drug screening data,that world might not be far off.In the near term,we hope to bring a TF targeting inhibitor into the clinic with positive Phase II data in t

312、riple negative breast cancer or high grade serous ovarian cancer.Partnerships:Weve worked with over 40 Pharma companies including Genentech,Pfizer,and J&J.Weve profiled their compounds and have sold back to them intelligence on how their drugs work,whether they are disease modifying or if they are e

313、ngaging pathways that might indicate downstream liabilities such as toxic side effects.Weve partnered with over 40 Pharma companies to screen their drugs for changes in the transcriptome.Our lead program has found novel chemistry to an extremely undruggable target that elicits a specific transcripto

314、mic effect.COMMERCIALVALIDATION50Decoding Bio Snapshot2023Joey Azofeifa,PhDAmit Mehta,MD,FRCPBill BarrettJoe McMahonJohn Dickson,PhDDave Morris,PhDCarter GrochalaDave SimpsonTaylor SchwerReo Yoo,MSJenna Rimel,PhDLeah Damon,PhDRiley Lundblad,MSCasey PhamChristy Tesman,PhDCole LiechtyLaura NorrisTim R

315、ead,PhDArdeshir Goliaei,PhDCEO&Co-founderBoard DirectorBuilders VCWSGR CounselIllumina,GRAILStrategic AdvisorKBI BiopharmaMedicinal ChemistryBristol-Myers SquibbPlatform AutomationGSKResearch AssociateLab Operations ManagerAdministrative ManagerResearch AssociateScientistScientistResearch AssociateA

316、utomation EngineerComputational BiologistResearch AssociateCOO&Co-founderCTO&Co-founderComputational ChemistThe TeamBoardCHARACTER BIOSCIENCESPrecision medicine for diseases of aging Founding year:2018Founders:Cheng Zhang,Marcel van der Brug,Nick JohnsonNumber of employees:21Location:South San Franc

317、isco,CA;Jersey City,NJ(USA)Last funding round:Series ATotal funding raised:$25MOur most advanced assets include a best-in-class complement inhibitor and first-in-class lipid modulator targeting distinct patient subgroups of dry Age-related Macular Degeneration(AMD),chosen due to the strength of clin

318、ical genomic evidence from our proprietary database.Our first indication targets the 200M patients worldwide with AMDWe are seeking to understand the molecular drivers and modifiers of disease progression endpoints,using human genetics and longitudinal clinical data as the foundation.These insights

319、lead to better drug target selection and patient stratification.Our approach builds a systematic understanding of patient molecular and clinical heterogeneity at the beginning of the drug discovery process.We are focused on:-Increasing likelihood of clinical development success and regulatory approv

320、al-Reducing cost and time of clinical development-Improving long-term patient outcomes and health economics with precision medicine.PRODUCT PIPELINE51TECHNOLOGYDecoding Bio Snapshot2023Market:We are targeting dry AMD as our first indication because(1)it is highly heritable(up to 70%)(2)there is high

321、 quality real-world clinical data(3)it addresses significant unmet needs with high patient heterogeneityVision:-Focus on multiple genetically defined chronic progressive indications-Molecular(re)classification of complex diseases to enable targeted therapies development-Multiple clinically validated

322、 drug assets and diagnosticsBusiness model:We follow a hybrid approach balancing the time efficiency and control of Internal Development on one hand,and the risk diversification and expertise sharing of Partnerships on the other handPartnerships:2019 research collaboration agreement Genentech,which

323、grants Genentech license to access a specific proprietary dataset built by Character Bio.Our other partnerships are undisclosed.COMMERCIAL52Decoding Bio Snapshot2023Cheng ZhangJoel DudleyPeter DiLauraVivek GaripalliCheng ZhangDavid Buchholz,PhDSumra FarooqTejasveetaNadkarniAnita WongYunju Yang,PhDIs

324、han Paranjpe,MDMarcel van der Brug,PhDNick JohnsonAndrea NeeranjanAmy ParkGeraldine TaverasWedward WeiAdria FrazierJonathan Gumu-cio,PhDFlorence LantanLaura MacriErin Burke,PhDRyan CampbellNick ChurchBrendan CurranErik Karrer,PhDBen Glicksberg,PhDMaria Avrutsky,PhDJulia BarberCEO,Co-FounderInnovatio

325、nEndeavorsLifeforce CapitalClover HealthCharacterSenior ScientistClinical Research AssociateSenior Research AssociateSenior ClinicalResearchCoordinatorHuman Genetics LeadData ScienceConsultantScientific Advisor,Co-FounderTechnology Advisor,Co-FounderSenior ClinicalResearch ManagerSenior ClinicalRese

326、archCoordinatorSenior ClinicalResearchCoordinatorMachine Learning EngineerResearch AssociateAssociate DirectorOperation AnalystSenior ClinicalResearch ManageStrategic Partnering LeadSenior ResearchAssociateAssociate ScientistSoftware EngineerSVP,Drug DiscoveryVP,Data ScienceSenior ScientistData Spec

327、ialistThe TeamBoardIMMUNAIHarness immunomics to improve therapeutics&the drug development processFounding year:2018Founders:Noam Solomon(CEO),Luis VolochScientific founders:Danny Wells,Ansuman Satpathy,Dan LittmanNumber of employees:150Location:New York,NY(HQ),Tel Aviv,Israel(office),Zurich,Switzerl

328、and(office),Prague,Czech Republic(office)Last funding round:Series BTotal funding raised:$300MImmunai has created a vertically integrated platform that leverages cutting-edge technology across multiple disciplines(i.e.single cell multi-omics,big data engineering,AI/ML,functional genomics and experim

329、ental immunology)to interrogate the immune system at scale.Why is this a breakthrough?We are unique in how we bring these technologies together were able to do single-cell immune profiling with different cells&reduce batch effects,while also storing and querying datapoints from thousands of patient

330、samples&millions of cells in a unified database,while providing the ML capabilities to mine insights from that database.We also use this to build superior in vitro model systems that better recapitulate human biology by matching cell states between preclinical models and patient data setsThe data-dr

331、iven insights we generate can accelerate drug development for immune-modulating medicines;drug development is currently too costly($2B),time-consuming(10+years)and failure-prone(95%failure rate)Our approach allows for“fast failures”early in the development process,greatly reducing cost&time investme

332、nt;we also believe we can improve success rates by identifying the right target,combo,patient population or indication for a specific product53TECHNOLOGYDecoding Bio Snapshot2023Market:We are mainly focused on partnering with the largest biopharma companies,both in the US and Europe.We do not have a

333、ny commercially available products.Vision:Successfully bridge the gap between causal immunology and translational-disease biology via single-cell immunology,engineering and AI to accelerate the development of next-gen therapeutics and improve patient outcomesBusiness Model:We are predominantly focused on delivering on our partnerships at this time doing so allows us to expand our capabilities/offe