TechBio Innovation: Latest Trends
TechBio Innovation: Latest Trends
AI/ML tools, startups, companies and investors
“Time is basically an illusion created by the mind to aid in our sense of temporal presence in the vast ocean of space. Without the neurons to create a virtual perception of the past and the future based on all our experiences, there is no actual existence of the past and the future. All that there is, is the present.”
By Abhijit Naskar, Love, God & Neurons: Memoir of a scientist who found himself by getting lost
Latest News 🍿
🧶 Insilico Medicine completes patient enrollment in its Phase IIa Study in China investigating INS018_055 for Idiopathic Pulmonary Fibrosis (IPF)
The primary objective is to evaluate the safety and tolerability of INS018_055 orally administered for up to 12 weeks in adult subjects with IPF compared to placebo. In addition, Insilico is preparing a Phase IIb proof-of-concept study to be initiated in 2025 to explore the efficacy and further safety of INS018_055.
🧶 AI Isn’t the Magic Bullet to Simplify Drug Discovery
Wet-lab biology and translational models are key to confirming AI-derived findings.
🧶 Formation Bio raises $372M to boost drug development with AI
Formation Bio, a startup focused on applying AI to drug development with backing from OpenAI CEO Sam Altman, it raised $372M in a Series D funding round led by Andreessen Horowitz with participation from drug maker Sanofi, Sequoia, Thrive, Emerson Collective, Lachy Groom, SV Angel Growth and FPV Ventures.
🧶 AI In Drug Discovery Market Size to Expand US$ 11.93 Bn by 2033
Report Highlights
North America region has accounted for a market share of around 56.18% in 2023.
The APAC market is expected to grow at a CAGR of 21.1% from 2024 to 2033.
By therapeutic area, the oncology segment has accounted for a market share of 21% in 2023.
Based on application, the drug optimization and repurposing segment has captured market share of 51% in 2023.
The US AI in drug discovery market size was valued at USD 670M in 2023.
🛑 SilicoGenesis
SilicoGenesis (2019, Belgium) is developing an AI-enabled in silico antibody engineering platform, providing state-of-the-art AI/ML technology integrated into a scalable cloud-based platform for: Precise 3D modeling of protein structures, Accurate prediction of paratope and epitope residues, Characterization and modeling of protein-protein interactions, Enhancement of binding affinity and mutagenesis studies, Cross-species reactivity analysis, Expert humanization of antibody candidates and Rigorous assessment of developability liabilities. Their cloud-based software platform for biologics design, discovery and optimization is called Eve.
They also offer Protkit, an Open Source Python library that can be used for a variety of tasks in computational biology and bioinformatics, focusing on structural bioinformatics, protein engineering and ML. Protkit allows you to download protein structures and sequences from a variety of trusted sources, such as RCSB PDB, Uniprot and SAbDab.
In particular, Protkit can be used for a variety of tasks in computational biology, such as reading and writing from popular data file formats, downloading data from popular protein databases, data structures for representing proteins and molecules, detecting and fixing anomalies in structures, calculating properties of proteins, running various external tools, featurization for ML, etc.
The company was founded by Lionel Bisschoff and Fred Senekal in 2021 in Johannesburg, South Africa and Ian Wilkinson is the scientific advisor. Ian is a protein biochemist with an interest in protein and antibody engineering that has previously acted as the CSO of Absolute Antibody (for antibody sequencing, engineering and recombinant manufacturing) and mAbsolve with a Fc silencing technology (which he co-founded) and is the founder of mAbvice consulting services.
Shortly after, SilicoGenesis’ headquarters were established in Leuven, Belgium. In 2022, their first academic collaboration was established on a project involving CART-T cell therapy, utilizing the Eve platform. In the first quarter of 2023, they begun collaborations with partners from around the globe including one of the largest pharmas based in Europe, on an affinity maturation project for a difficult cytokine target:
⏹️ Given the sequence of the antibody candidate, they predicted its structure and PPI interaction with the target antigen. They successfully modeled the three-dimensional structure of the complex and used this highly accurate model to perform in silico saturation mutagenesis. They were able to correctly identify all of the top affinity-enhancing mutations in the CDR regions confirmed by in vitro saturation mutagenesis using ELISA and SPR.
⏹️ They performed rapid in silico affinity maturation of a lead candidate in less than a month.
Recently, SilicoGenesis began a collaboration with the Laboratory for Thrombosis Research (KU Leuven) and PharmAbs (KU Leuven Antibody Centre). The project involves in silico antibody engineering and optimization as well as in vitro and in vivo testing and validation. The goal of the project is to determine if the engineered lead molecule has improved developability characteristics and preserves binding affinity and species cross-reactivity.
🥎 FabricNano Limited
FabricNano (2018) is a biotechnology company based in London with the mission to transform industrial chemical processes using cell-free biomanufacturing, with an advanced, flexible and easily-scalable biocatalyst platform. They develop and sell biocatalysts for high volume industrial applications, utilizing a data-driven approach for enzyme and Immobilization Engineering™. Their clients range from startups to international clients like Sumitomo Chemical Company (FabricNano and Sumitomo Chemical America partner to develop the next generation of cell-free bio-manufacturing), Ginkgo Bioworks (Ginkgo Bioworks partners withFabricNano to unleash Immobilization Engineering™ for a wide range of enzyme discovery projects), ALMAC and many more.
The FabricNano process starts with novel Immobilization Engineering™ for enzyme stabilization, followed by budget-conscious protein engineering and process engineering to reach the client’s targets for commercialization of the new biochemical production process.
In other words, FabricNano collects all the enzymatic components necessary for biomanufacturing and co-locates them on an artificial surface. The company has tried everything from a nano-wafer fabric made of DNA, to more industrially available materials such as common solid carriers, coffee grinds and even simple rocks (FabricNano is pioneering sustainable chemical production with cell-free biocatalysis). Once the enzymes are durable and stable they can be used to manufacture commodity chemicals like bio-plastics, bio-fuels and certain types of antibiotics that are made using enzymes. For example, the Epoxy E0001-0488-01 with Penicillin Acylase.
Furthermore, these biocatalysts 🔩 are what is known as “drop-in,” so they can just go into pre-existing industry equipment and know-how (such as packed bed or continuous stirred tank reactors).
The company is backed by Atomico and Hoxton VC (as well as prominent angel investors) and has raised a total funding of $22.6M over 2 rounds from 12 investors. Founder and CEO at FabricNano is Grant Aarons.
🔊 OMNIVORE 👓
Omnivore is the free, open source, read-it-later app for serious readers for saving articles, newsletters and documents, and reading them later—focused and distraction free. With Omnivore you can add notes and highlights, organize your reading list the way you want and sync it across all your devices.
⚾ cfdx
cfdx (2023, London, UK) is an AI neuroscience company, currently in stealth mode, co-founded by Hannah Thompson, PhD and Jonathan Wan.
cfdx aims to apply the liquid biopsy techniques used in early cancer detection to neurodegenerative diseases, enabling diagnosis decades before symptoms appear.
🪀 StoneWise (Beijing StoneWise Technology Co, Ltd)
StoneWise (2018, Beijing, China) (望石智慧) utilizes AI to enable knowledge mining, molecule generation and property prediction allowing researchers to build knowledge graphs of scientific literature, predict molecular properties, design novel molecules and perform retro-synthetic analysis. StoneWise has set up an efficient and practical AI-based drug design platform, StoneMIND® (Master of Intelligent Novel Design), which provides a unified SaaS platform for information extraction, knowledge mining and molecular design. In particular:
🏗️ StoneMIND Collector (Information Extraction System): Is an AI enabled software tool capable of quickly extracting chemical structures from patent documents, literature and various pictures, and users can easily edit physical and chemical property data of the extracted structures. The obtained structural information and analysis results can be exported to various computer-readable formats at any time (SDF/MOL/SMILES/XLS/PNG).
🏗️ StoneMIND Designer (Molecular Design System): Is a multi-layer model for molecule design and optimization based on ligands, receptors and fragments, empowered by an intricate combination of AI, domain knowledge and a massive database. With the support of high-throughput screening of structures, is making it possible for researchers to discover quality molecules even faster.
An ultra-high throughput molecule screening system (Ultra-HTS) allows virtual high-throughput screening at billions of magnitude, that is combined with multi-dimensional molecular generation and optimization systems to enable faster and better drug discovery.
🏗️ StoneMIND Inspirer: Is an auxiliary decision-making tool of SAR visualization analysis for R-group optimization. It aims to help researchers conduct statistical analysis of chemical and biological data, quickly carry out SAR research, and give reasonable suggestions on fragment alteration. It provides one-stop assistance for structural modification and alteration of small molecules, so as to bring molecular design into the era of rational design.
🏗️ KDD (Knowledge Discovery in Database): System of massive data annotation, relations optimization and analytic tools enable easier collection, categorization and archiving.
Founder and CEO of StoneWise is Zhou Jielong. In 2018, Zhou Jielong, a key force behind the evolution of Baidu’s search engine, established StoneWise to aid in the discovery of small-molecule drugs (AI drug discovery booms in China). Zhou Jielong (Master Degree of Artificial Intelligence, Beijing Institute of Technology) was Baidu’s former Principal Architect, and was responsible for application of ML technologies into Baidu’s core business, including core page ranking, anti-spam, cloud-based voice search and image search algorithms. He led the team to re-frame Baidu’s search engine with ML algorithms, and made innovative efforts in a few AI fields, including AI system, AI expandability and AI robustness. In 2013, he led his team to successfully apply DL to search engines for the first time in the world.
In 2021, StoneWise completed its series B and series B+ funding rounds raising $100M combined.
🧮 Krarup Analytics 🧮
A single person company based in Copenhagen, Denmark, for linking analytics to your future competitiveness: Krarup Analytics
⚽ Recursion Pharmaceuticals Inc (NASDAQ: RXRX)
Recursion (2013, Utah US) is a leader in digital biology, and has built the world’s most advanced ultra-high throughput wet-lab and ML platform. Their ability to generate proprietary, high-dimensional, multi-modal and relatable datasets of human cellular biology at massive scale, and apply advanced ML approaches to reveal novel biological relationships, has resulted in a proven, target-agnostic drug-discovery engine. Recursion Pharmaceuticals Inc (RXRX) went public in April 2021.
Recursion Pharmaceutical has at its headquarters clusters of robots that treat millions of cells per week with drugs, stain them with six dyes and then take pictures to capture and quantify as many morphological features as they can. By pushing these data through a ML pipeline, they hope to find relationships that are invisible to the human and to tease out clusters of effects that can guide their drug discovery.
⏺ In 2020, a strategic collaboration was announced that will leverage Recursion’s purpose-built AI-guided drug discovery platform and Bayer’s small molecule compound library and deep scientific expertise to discover and develop new treatments for fibrotic diseases of the lung, kidney, heart and more. Under the terms of the agreement, the parties can initiate more than 10 programs with possible development and commercial milestone payments of more than $100M per program plus royalties on future sales. In addition to the $50M equity investment, Recursion will receive an upfront payment of $30M.
On November 10, 2023, Bayer and Recursion expanded their oncology research partnership and may launch up to seven oncology programmes.
Recursion is entitled to receive development and commercial milestone payments of $1.5bn, along with royalty payments on future net product sales.
Bayer will be the first beta-user of their LOWE LLM-orchestrated workflow software, which will be integrated across the collaboration and offer a more exploratory, and intuitive research environment for scientists on both sides.
LOWE is an LLM agent that represents the next evolution of the Recursion OS. LOWE supports drug discovery programs by orchestrating complex workflows. These workflows chain together a variety of steps and tools, from finding significant relationships within Recursion’s Maps of Biology and Chemistry to generating novel compounds and scheduling them for synthesis and experimentation. Through its natural language interface and interactive graphics, LOWE puts state-of-the-art AI tools into the hands of every drug discovery scientist at Recursion in a simple and scalable way.
Additional updates pertaining to the Bayer partnership include:
Their first joint oncology project is now expected to advance rapidly towards Lead Series nomination; and
They are on track to complete 25 unique multi-modal data packages that they expect to deliver in the third quarter of 2024.
⏺ On September 2020, it was announced by the company an oversubscribed series D funding round of $239M. Recursion has raised a total of $665.4M.
⏺ On December 7, 2021, Recursion announced a transformational collaboration with Roche and Genentech, in order for Recursion to work with both Roche and Genentech's R&D units to leverage technology-enabled drug discovery through the Recursion Operating System (OS). Under the terms of the agreement, Recursion will receive an upfront payment of $150M while Roche and Genentech (combined) may initiate up to 40 programs, each of which, if successfully developed and commercialized, could yield more than $300M in development, commercialization and net sales milestones for Recursion, as well as tiered royalties on net sales.
On October 2, 2023, just over one and a half years into the exciting collaboration with Roche and Genentech, Recursion announced that they have reached the first milestone: after building fit-for-purpose oncology maps for their partner spanning whole-genome arrayed CRISPR knockouts and hundreds of thousands of small molecules, they have identified and validated the first hit series for this particular disease, and Roche has exercised the Small Molecule Validation Program Option.
⏺ On June 12, 2023, Recursion acquired the two emerging Canadian AI drug-discovery firms: Cyclica and Valence. Subsequently, Recursion announced a collaboration and a $50M investment from NVIDIA. Then Recursion launched Valence Labs, formerly Valence Discovery, a company with roots at Mila and mentorship from Yoshua Bengio (a Canadian computer scientist, most noted for his work on AI networks and DL), dedicated to advancing DL in drug discovery, delivering impactful research and transformative technology and embracing open-source and open-science knowledge sharing with the ML community.
⏺ On January 8, 2024, Nvidia Corp. (NVDA) doubled down on AI powered drug discovery and development, announcing an expanded partnerships with Amgen Inc (AMGN) and Recursion Pharmaceuticals Inc (Nvidia dives deeper into AI drug development with Amgen, Recursion partnerships). Amgen’s subsidiary deCODE Genetics is building out a supercomputer to create genomics “foundation models”—models trained on massive datasets to tackle a variety of jobs—to fuel drug discovery. deCode will power its new genomics foundation models with Nvidia’s supercomputer and BioNeMo generative AI platform. Recursion is also joining the party and will be the first third-party addition to the BioNeMo platform, adding its Phenom-Beta program for wider use.
⏺ On May 12, 2024, it was announced that NVIDIA bought $70M of shares in RXRX 0.00%↑. Recursion Pharmaceuticals collaborates with NVIDIA to accelerate drug discovery using AI technology. Their newest supercomputer, BioHive-2, powered by 504 NVIDIA H100 Tensor Core GPUs, delivers 2 exaflops of AI performance—nearly 5x faster than their previous system (Recursion Pharmaceuticals is at the forefront of AI-driven drug discovery!).
Recursion’s BioHive-2 is the largest system in the pharmaceutical industry. It’s located at Recursion’s headquarters in Salt Lake City.
This supercomputer 🦸♂️ ranks No. 35 on the latest TOP500 list of the world’s fastest supercomputers.
Recursion’s integrated Recursion Operating System creates a closed-loop system combining proprietary in-house data generation and advanced computational tools to generate novel insights to initiate or accelerate therapeutic programs. Afterwards, their in silico predictions are validated in their own wet laboratories, and repeated, creating a mutually reinforcing cycle of learning. So far, they have something like ~19 petabytes of proprietary high-dimensional data and the company has the following clinical trials:
🧑🏭 REC-994 in phase 2 for
Cerebral Cavernous Malformation (CCM). Phase 2 SYCAMORE clinical trial is a randomized, double-blind, placebo-controlled, safety, tolerability and exploratory efficacy study in participants with CCM.
Topline Phase 2 data readout in September 2024.
👨🏭 REC-2282 in phase 2 for
Neurofibromatosis Type 2 (NF2). Phase 2/3 POPLAR clinical trial is a randomized, two part study in participants with progressive NF2-mutated meningiomas. They expect to share data in 2024.
REC-2282 (a pan-histone deacetylase inhibitor) was discovered with the promise for accelerated development. Recursion in-licensed the compound from Ohio State University to be developed for NF2, and their first wholly owned subsidiary company CereXis, Inc is currently preparing REC-2282 for Phase 2 clinical trials.
Preliminary Phase 2 data readout in the fourth quarter of 2024.
🧑🏭 REC-4881 in phase 2 for
Familial Adenomatous Polyposis (FAP). Phase 1b/2 TUPELO clinical trial is an open label, multicenter, two part study in participants with FAP.
Preliminary Phase 2 data readout in the first half of 2025.
AXIN1 or APC Mutant Cancers. Phase 2 LILAC clinical trial is an open label, multicenter study in participants with unresectable, locally advanced or metastatic cancer with AXIN1 or APC mutations.
Preliminary Phase 2 data readout in the first half of 2025.
🧑🏭 REC-3964 in phase 1 for
Clostridioides difficile Infection. In a Phase 1 healthy volunteer study to evaluate the safety, tolerability, and PK of REC-3964 at increasing oral doses in comparison with placebo, REC-3964 was safe and well tolerated and there were no serious adverse events, or deaths.
Phase 2 study initiation in the fourth quarter of 2024 and preliminary data readout by the end of 2025.
👨🏭 RBM39, a novel CDK12-adjacent target identified by the Recursion OS, for HR-Proficient Ovarian Cancers and Other Solid Tumors. As a result of their collaboration with Tempus, they are leveraging genomic data across all tumor types to identify clinical biomarkers for patient expansion.
IND submission in the third quarter of 2024, Phase 1/2 initiation in the fourth quarter of 2024 and Phase 1 dose-escalation data readout by the end of 2025.
👩🏭 Undisclosed Indication in Fibrosis (Target Epsilon) from the fibrosis collaboration with Bayer which is now entering initial investigational new drug (IND) enabling studies.
IND submission in early 2025 and Phase 1 healthy volunteer study data readout by the end of 2025.
👨🏭 Dozens of internal and partner programs in early stages with the first LLM and causal model driven programs entering the Recursion pipeline.
🍄 MOLLEO 🍄
Efficient Evolutionary Search Over Chemical Space with Large Language Models
Molecular discovery, when formulated as an optimization problem, presents significant computational challenges because optimization objectives can be non-differentiable.
Evolutionary Algorithms (EAs), often used to optimize black-box objectives in molecular discovery, traverse chemical space by performing random mutations and crossovers, leading to a large number of expensive objective evaluations.
In this work, a group of researchers ameliorated this shortcoming by incorporating chemistry-aware Large Language Models (LLMs) into EAs. Namely, they redesigned crossover and mutation operations in EAs using LLMs trained on large corpora of chemical information.
They performed extensive empirical studies on both commercial and open-source models on multiple tasks involving property optimization, molecular rediscovery, and structure-based drug design, demonstrating that the joint usage of LLMs with EAs yields superior performance over all baseline models across single- and multi-objective settings. Finally, they demonstrated that their algorithm improves both the quality of the final solution and convergence speed, thereby reducing the number of required objective evaluations.
🏈 Tierra Biosciences
Tierra Biosciences (2015, US) is a synthetic biology company that accelerates the pace of discovery and enables the next generation of bio-based materials, with its AI-driven Tierra Protein Platform, that couples high-throughput protein manufacturing with the simplicity of e-commerce to make ordering custom proteins easier than ever before.
Advances in DNA synthesis, DNA sequencing and de novo protein design have made exploration of diverse sets of sequences easier. But the challenge has now moved to protein synthesis and the solution is the cell-free expression Tierra Protein Platform, where you can:
Input your digital protein sequences and get AI-driven computational insights into synthesis profiles and risks.
Then receive physical, testable proteins that have gone through extensive experimental validation and QA/QC. All through a single digital portal.
Tierra’s proprietary cell-free expression technology utilizes the components of a living cell, but not the cell itself, since engineering and manipulating cells can be a challenge. Then by combining automation, computational analysis and high-throughput cell-free expression, is synthesizing proteins from diverse sources.
By providing high throughput rapid protein synthesis from digital protein sequences, Tierra allows you to move quickly, efficiently, and cost-effectively to your downstream assays to screen and identify hits.
Finally, you can get your custom, purified µg- and mg-scale proteins in as little as 3 weeks.
Tierra’s technology was originally developed at Caltech. Co-founders are Zachary Sun and George Church, one of the living legends of modern genetics.
With applications in biopharma (Antibody engineering, Difficult-to-express proteins and Drug Targets) and synthetic biology (Computationally designed proteins en masse, Protein engineering, Toxic Proteins, Pathway and metabolic engineering and Metagenomic Screening), Tierra Biosciences raised in March 2024 a $11.4M series A to expand its designer protein-to-order platform.