Table of Contents:
Artificial intelligence (AI) in drug discovery yields breakthroughs
The abundance of venture capital, major funding rounds
IPOs of AI Pharma Companies in 2021
New AI-driven biotech startups founded in 2020
Notable AI-focused collaborations involving “big pharma” players
AI adoption by the contract research organizations (CROs)
According to GlobalData’s report ‘Smart Pharma’, 28 percent of all the surveyed companies plan on using artificial intelligence (AI) and big data technologies to optimize drug discovery and development processes in the next two years. Furthermore, 32 percent of respondents would be employing big data analytics and predictive technologies to streamline sales and marketing efforts. This indicates growing practical importance of AI in both scientific and operational aspects of the pharmaceutical business.
Artificial intelligence in a nutshell is a field of science concerned with creating intelligence agents -- systems capable of taking in data from the surrounding world, processing it in some intelligent way, and outputting meaningful results. A powerful sub-field of artificial intelligence is machine learning (ML) -- a broad scope of algorithms capable of improving its prediction accuracy with each new iteration, or with more data available for training. Both artificial intelligence and machine learning (ML) are old concepts, known for decades, but the actual AI revolution in many industries and in pharma, in particular, has been enabled by a specific type of machine learning -- deep learning (DL). This family of AI models is based on the architecture of neural networks, and owing to the unprecedented ability to learn sophisticated data representations -- especially from multimodal data -- deep learning algorithms can provide insights into data unattainable by any other statistical methods and models. It is deep learning that drives all the buzz around the area of artificial intelligence (sometimes very well deserved, other times -- overhyped), and it is deep learning that is set to disrupt the way drug discovery research is done.
According to BiopharmaTrend’s interactive report “The Landscape of Artificial Intelligence (AI) In Pharmaceutical R&D”, various machine learning and deep learning models can be applied to improve a wide range of research tasks in drug discovery, including de novo drug design, structure-activity relationship (SAR) prediction; improvement in 3D protein structure simulation; biomarker discovery, patient stratification, predicting drug responses; analysis of high-content screening results (e.g. cell images) in real-time, correlation to other types of data; improving drug repurposing programs, streamlining clinical trial operations, and much more.
AI in drug discovery yields breakthroughs
While there is an abundance of proof-of-concept studies in applying artificial intelligence in drug discovery, several success stories are especially illustrative.
One such case is Insilico Medicine, which recently reported a new drug candidate for kidney fibrosis, having repeated their previous success in identifying a drug candidate for idiopathic Pulmonary Fibrosis (IPF) -- all using their drug design system Chemistry42, based on generative adversarial neural networks (GANs). Both their successes were enabled by a more important discovery -- that of a novel pan-fibrotic target, revealed by their target discovery platform PandaOmics. The entire process from the initial target hypothesis to the preclinical drug candidate nomination took under 18 months and a fraction of the cost of “traditional drug discovery”.
Another prominent example is DeepMinds AlphaFold2 -- the AI system for predicting the 3D shape of proteins based on their primary amino acid sequence. It is believed to have changed the paradigm of modern structural biology -- just recently, the company released predicted structures of as much as 350.000 proteins, including nearly every protein expressed in the human body.
The AlphaFold Protein Structure Database of the human proteome was published in Nature and is made openly available for the drug discovery community.
Next, in April 2021, Evotec announced a phase 1 clinical trial on a new anticancer molecule, created in partnership with Exscientia -- a UK-based drug discovery company, specializing in AI-driven small molecule design. It took Exscientia around 8 months to come up with an A2 receptor antagonist efficient in helping T cells fight solid tumors -- using their AI platform Centaur Chemist. In 2020, Exscientia in partnership with Sumitomo Dainippon Pharma developed a selective serotonin reuptake inhibitor (SSRI) designed to treat obsessive-compulsive disorder (OCD), which has also advanced to a clinical state.
The growing ecosystem of organizations developing or adopting pharmaceutical artificial intelligence is represented by more than 280 drug discovery companies in the relatively early stage of their development (most founded within the last 5 years). Several of them managed to raise hundreds of millions in venture capital or even went on IPOs just recently.
Besides, almost all leading pharmaceutical organizations (‘big pharma’) and largest contract research organizations (CROs) are intensively developing (or at least exploring) in-house AI-driven systems and data-centric workflows in R&D and operations, and actively partnering with AI-vendors and R&D platform providers. A searchable database of such collaborations and financial rounds is available in our report “The Landscape of Artificial Intelligence (AI) In Pharmaceutical R&D”, however, below we will list some notable examples to illustrate the industry developments.
One important trend now is the expansion of leading non-pharmaceutical tech corporations, such as Google, Microsoft, Tencent, Baidu, Nvidia, Intel, etc, into the pharmaceutical research and biotech industry -- something that is largely enabled by their cutting-edge technologies in AI and big data. Besides, a wide array of technology providers of a smaller scale are increasingly present in the pharmaceutical market -- from big data companies, like 23andMe, to quantum computer developers, like Zapata Computing, and autonomous lab developers, like Arctoris and Strateos.
A growing ecosystem of advanced technology providers operating in the pharmaceutical industry and biotech is profiled in a report by Deep Pharma Intelligence “Landscape of Advanced Technology Companies in Pharmaceutical Industry Q2 2021”.
Now, let’s review some of the notable industry developments related to the use of various artificial intelligence technologies, such as deep neural networks and natural language processing (NLP) tools being applied in drug discovery and clinical research. The focus here is on major investment rounds, illustrative R&D partnerships between AI-driven companies and “traditional” pharmaceutical organizations. We will also review new AI-driven biotech startups founded in 2021. This is not meant to be a comprehensive overview, rather – an illustrative list of developments and reflections on where things are moving forward. For comprehensive coverage of the pharmaceutical AI ecosystem, check our interactive report “The Landscape of Artificial Intelligence (AI) In Pharmaceutical R&D”.
The abundance of venture capital and new IPOs in the pharma AI space
The last year has been a record-breaking year for biotech funding in general, with lots of venture capital activity, a kaleidoscope of IPOs, and substantial growth in biotech indices. The sub-area of the pharmaceutical AI also received increased attention by all kinds of investors, with almost $4 billion raised collectively by early-stage AI-driven drug discovery companies in private rounds and public market launches. Notably, a comparable amount of money was raised by such companies in new rounds, later-stage deals, and new IPOs in just the first half of 2021, meaning the funding dynamics is accelerating.
Cellarity creates an intelligent system of cellular behaviors analysis, called the Cellularity Platform. The platform models cellular behaviors/misbehaviors in different medical conditions and generates drug candidates that potentially will affect these behavioral patterns. It is a new approach to drug development based on artificial intelligence technology and computational disease modeling. Cellarity was founded in 2017 and in February 2021 it raised $123 million in Round B, bringing the total company’s capital up to $173 million.
BigHat Biosciences is a California-based company that develops next-generation antibody therapeutics. It leverages machine learning platforms NETMAPPR and CombiGEM to learn complex gene interactions and uses advanced wet lab practice to verify and further develop therapies based on acquired knowledge. BigHat’s AI/ML technologies drive the discovery of novel therapies, analyzing what kind of antibody mutations affect their molecular properties. In February 2021 it banked $19 million in Round A, led by 4 investors.
In March 2021 Insitro banked $400 million of investments from 14 different investors worldwide. Insitro starts the drug discovery process with AI-driven analysis – extracting valuable insights from phenotypic, genetic, and clinical data. Next, ML algorithms and experts combine their efforts to build disease models on induced pluripotent stem cells. Big amounts of high-quality data are generated and interpreted by AI technologies to unveil novel therapeutic targets.
2021 has been an exciting year for Exscientia, which raised $325 million in two successive Series C and D rounds, and complimented them with a grant from the Bill & Melinda Gates Foundation worth $1.5 million. The company’s pipeline contains 20+ candidates developed with the help of AI algorithms, with 3 of them already in the preclinical stage and 1 in the Phase 1 trial.
Dyno Therapeutics works in the area of progressive gene therapies development, with the help of AI it develops a platform that will create gene vectors suitable for a broader pool of patients and that can target more diseases than they do today. Its CapsidMap™ Platform is used to optimize Adeno-associated virus (AAV) capsids and improve their targeting potential, payload size, reduce immunoreactivity caused by vectors and advance manufacturing properties. Since 2018, Dyno raised $109 million with the latest round in May 2021. The company already signed 3 significant R&D deals – with Roche, Sarepta and Novartis.
A patented machine learning-based platform is being created by Engine Biosciences, the platform will be used to analyze complex gene interactions, identify genetic errors that cause health issues and discover CRISPR breaks with healing potential. In addition, Engine has its own discovery and preclinical pipelines in the area of precision cancer medicine and collaborates with partners to develop drugs in different spheres. In May, Engine raised $43 million from Polaris Partners, EDBI, Baidu Ventures and other organizations.
Enveda offers an end-to-end Drug Discovery platform with integrated multi-omics datasets for the development of therapeutics for inflammatory, fibrotic, and neurological diseases. The discovery platform exploits MOA-driven hypotheses in data streams integration and identification of target candidates, cloud-scale dereplication and bioactive lead identification techniques. Combined together, they help Enveda in the discovery of the first-in-class therapeutics. Over 2 years of the company's existence, $55 million was raised, with $51 million received in June 2021 from Village Global, Two Sigma Ventures and Lux Capital.
Insilico Medicine is a Hong-Kong based drug discovery company that created an end-to-end AI drug discovery platform, including three main components: the target discovery system PandaOmics, the drug design module Chemistry42, and clinical trial prediction system InClinico. PandaOmics is used to identify novel targets by analyzing big masses of multi-omics data within a few clicks, Chemistry42 generates lead-like molecules with de novo machine learning drug design and the InClinico platform discovers clinical trial weak points and predicts their success rate. In June 2021 Insilico attracted $255 million of Series C funding from a list of prominent investors.
The list of Insilico’s partners includes Prizer, Boehringer Ingelheim, Astellas, Taisho and Beijing Tide Pharmaceutical.
IPOs of AI Pharma Companies in 2021
2020 and the beginning of 2021 has been a lucrative period for biotech IPOs in general, with several IPOs occurring in the segment of AI-driven biotechs.
One example is Adagene, a clinical-stage biotech company that specializes in AI-powered antibody drug discovery and occupies a niche of immuno-oncology therapies, in January 2021 claimed its plans to go public at $125 million. Next month it closed IPO with an unannounced amount, however, now their market cap reached $837 million.
Another startup, operating in the field of AI-driven immunotherapies development, is Evaxion Biotech, based in Denmark. Evaxion closed an IPO in February, getting $30 million.
Notably, special purpose acquisition companies (SPACs) gained traction in the pharmaceutical industry as a vehicle to bring biotech companies to public markets. One example is AI-driven personalized medicine biotech Valo Health, which went public in June 2021 -- via a $2.8 billion SPAC deal with Khosla Ventures.
New AI-driven biotech startups founded in 2020
Celeris Therapeutics is an Austrian drug discovery company working in the field of undruggable pathogenic protein targets with a computer-based deep learning approach. Celeris One -- the AI platform developed by Celeris Therapeutics, predicts the potential use of degrader technologies like PROTACs and Molecular Glues by analyzing biomolecular interactions between a target and a degrader. Celeris harnesses degrader technologies for treating Alzheimer’s disease and a broad range of cancers, including prostate and breast cancer.
Founded in January 2021, the biotech has already gathered $1.1 million in two Seed Rounds that took place in March and July.
Omnia Biosystems’ is a London-based biotech startup whose goal is to create a smart peptide therapeutics discovery system driven by neural networks to optimize time & financial spends. Inspired by computer-aided drug design, Omnia develops AI-based molecular drug design and currently has four candidates in the pipeline – three of them against various types of cancer and one designed to activate T cells.
NonExomics apply machine learning technology to discover drugs for diseases arising specifically from non-exom issues. The company integrates all kinds of omics data – genomics, transcriptomics, proteomics – to predict nonexomic proteins and identify disease-causing targets. NonExomics partners with other biopharma organizations to design biomarker assays, identify biomarker candidates, therapeutic targets and drugs.
‘Big pharma’ explores AI capabilities via R&D partnerships
Based on the agreement between AstraZeneca and BenevolentAI in 2019, AstraZeneca chose a new chronic kidney disease (CKD) molecular target to work on in January 2021. In the partnership, AstraZeneca is responsible for the scientific expertise and BenevolentAI produces extensive datasets with the help of its proprietary AI-driven platform. The novel CKD target will be the first AI-generated target to appear in AstraZeneca's portfolio. As an active participant of AI adoption in the pharma space, this year AstraZeneca also partnered with NVIDIA and the University of Florida on a new innovative artificial intelligence project. NVIDIA and AstraZeneca revealed their innovative AI drug discovery model called MegaMoIBART designed for molecular optimization, reaction prediction, and de novo molecule generation.
In March 2021, Pfizer entered a partnership with Iktos to deploy Iktos’s AI-based de novo drug design software Makya™ and generate small molecules for Pfizer’s discovery programs. This approach will replace lengthy library screening procedures and accelerate drug discovery for the pharma giant.
In April 2021 AbCellera entered a multi-year collaboration with Gilead Sciences to use AbCellera’s humanized mouse drug discovery platforms – the Trianni Mouse® platform and the OrthoMab protein engineering platform. Under agreements of the partnership, AbCellera will work on panels of antibodies against 8 targets chosen by Gilead with multiple applications. AbCellera and Gilead have already worked together in 2019 with the successful completion of the first program dedicated to fighting infectious diseases.
Another new Gilead Sciences collaborator is Gritstone Oncology – a company specializing in AI development of immunotherapies against various cancer types and infectious diseases. The partnership is created to discover treatments for the human immunodeficiency virus (HIV). Gritstone will provide services in the form of its proprietary prime-boost vaccine platform applied to antigens from Gilead, and Gilead will perform Phase 1 clinical trials. Overall, Gilead pays an upfront cash installment of $30 million and $30 million more as an equity investment, as well as $725 million in regulatory and commercial milestones achieved and royalties on net sales.
In May 2021 UK-based AI drug discovery leader Exscientia signed a deal with Bristol Myers Squibb to discover and develop novel small molecule drugs. Using Exscientia’s platform, new treatments targeted for multiple medical conditions, including oncological and autoimmune disorders, will be generated. An upfront payment from Bristol Myers Squibb reached $50 million and near-term success of treatments will bring $125 million more, with the final goal of up to $1.2 billion in case of clinical success, FDA approval, and commercialization.
In the same month, Israeli technological startup CytoReason teamed up with Swiss company Ferring Pharmaceuticals to discover treatments for inflammatory bowel disease (IBD). CytoReason’s AI-based computational model of the human body will serve as an accelerator for the drug discovery process, combined with the medical expertise Ferring gains. The outcome of the project will be presented in a form of novel therapeutic targets for further drug development. CytoReason’s technology is already leveraged by big pharma representatives like Pfizer, Roche, Sanofi, and GSK.
In July, Japanese company Sosei Group Corporation joined up with InveniAI to work on the new R&D program together. The collaboration will mobilize InveniAI’s AlphaMeld® platform to tackle the problem of adverse reactions to existing immunotherapies and immune diseases. The R&D research will reveal the role of G-protein coupled receptors (GPCRs) in immunomodulatory pathways and discover new targets to affect it.
AI adoption by the contract research organizations (CROs)
Most CROs are focused on late-stage drug development and clinical trials, however, nowadays more corporations are adopting early discovery stages as well, offering end-to-end therapeutics development. To stay competitive on the market and manage resources intelligently, many CROs partner or acquire AI startups and incorporate their know-how and services into existing value offerings.
In one example, Charles River Labs inked a strategic partnership with Valence Discovery to provide Charles’s clients with AI technology for molecular property prediction, generative chemistry, and multiparameter optimization. Valence’s platform is based on the research conducted at Mila, the world’s top deep learning research institute, and allows de novo design of small molecules in new chemical spaces.
In the interview, the executive director of the Institute of Data Science of IQVIA Murray Aitken said that artificial intelligence, machine learning, digital, telehealth and other key components of technological innovation were essential during pandemics for CROs, as clinical trials were mostly decentralized. Still, using progressive approaches helped CROs to maintain data quality and patient safety in these restricting conditions.
Global leader of outsourced drug development services ICON decided to acquire PRA Health Sciences Inc with a total estimated value in cash and stocks – $12 billion. The companies in cooperation will provide the market with decentralized and hybrid trial solutions, applying PRA’s expertise in artificial intelligence.