Weekly Tech+Bio Highlights #6
ALSO: Reinforcement Learning + LLMs = Next Gen 3D Molecular Design; How Teva Uses AI Today; Life Science E-commerce Embrases Machine Learning; Gene Editing Breakthrough: Going Beyound CRISPR
Hi! I am Andrii Buvailo, and this is my weekly newsletter, ‘Where Tech Meets Bio,’ where I talk about technologies, breakthroughs, and great companies moving the biopharma industry forward.
If you've received it, then you either subscribed or someone forwarded it to you. If the latter is the case, subscribe by pressing this button:
Now, let’s get to this week’s topics!
News Highlights
🚀 Ex-Meta scientists launch EvolutionaryScale with $142M seed round, supported by Amazon and Nvidia, to develop AI models for generating novel proteins for scientific research. The startup aims to revolutionize drug discovery and materials science with its ESM3 model, available for non-commercial use.
🧬 Researchers at the Arc Institute in California unveil the "bridge editing" DNA mechanism, enabling larger and more precise genomic modifications than CRISPR-Cas9, potentially revolutionizing genome engineering through scarless and extensive DNA alterations.
💰 Formation Bio secures $372MM in Series D funding led by a16z and Sanofi, aimed at expanding its AI-driven drug development platform and acquiring new drug candidates.
🤝 Helix and Recursion Pharmaceuticals announce a multi-year agreement to use Helix’s large-scale clinico-genomic data to train AI models for accelerating drug discovery and developing precision treatments, leveraging diverse multimodal datasets.
🚀 Boston-based Exsilio Therapeutics emerges from stealth with $82M in Series A funding to tackle gene therapy challenges using mRNA technology for whole gene delivery, aiming to create repeat-dosable genomic medicines for complex diseases.
🛒 Scientist.com, a leading R&D outsourcing marketplace, launches Elisa, an LLM-powered research assistant for the pharmaceutical and biotechnology industries, designed to enhance drug research by providing instant access to marketplace information and understanding complex research needs.
💉 Alnylam Pharmaceuticals announced that its heart drug, vutrisiran, showed positive Phase 3 results, reducing the risk of death and recurrent cardiovascular complications by 28% compared to placebo.
📈 New York-based LB Pharmaceuticals is set to file for an IPO soon, driven by rising investor interest in neuroscience. The company aims to advance LB-102, an improved version of the schizophrenia drug amisulpride, into Phase 3 trials by 2025 with an FDA approval target of 2028.
💰 Amylyx acquired rights to Eiger Biopharmaceutical’s GLP-1 receptor antagonist, avexitide, for $35.1 million, marking a strategic move into the metabolic diseases market following the failure of its ALS therapy, Relyvrio.
Unlike mainstream therapies that address hyperglycemia, diabetes, and obesity, such as Novo Nordisk’s Ozempic (semaglutide) and Eli Lilly’s Mounjaro (tirzepatide), avexitide is intended to treat patients with low blood sugar.
In the latest Thursday premium newsletter I’ve explored the market dynamics, companies, and pipelines of weight-loss drugs.
🔍 BenevolentAI and AstraZeneca's collaboration progresses with the addition of a novel target for systemic lupus erythematosus (SLE) to AstraZeneca's discovery portfolio, showcasing the success of BenevolentAI’s AI-driven drug discovery platform.
🔬 Proscia partners with Nucleai to integrate AI-driven predictive biomarkers into Proscia's Concentriq® platform, enhancing precision medicine by providing advanced diagnostics and optimizing clinical trial execution for better patient care.
🔬 ImmunoPrecise Antibodies' subsidiary BioStrand and PGxAI have partnered to develop an AI model for pharmacogenomics recommendations using BioStrand's LENSai™, enhancing precision medicine through personalized drug response predictions. This collaboration aims to transform medication prescription practices globally by integrating advanced AI and genomic data analysis.
🤝 Cradle has partnered with Google Cloud to enhance security for its AI-driven protein engineering platform, ensuring robust protection for sensitive research data and intellectual property.
💰 Curie.Bio, a venture capital firm co-founded by Alexis Borisy, raised $380 million for early-stage biotech investments, bringing its total to nearly $1 billion. The funds will support series A rounds for clinical proof-of-concept studies, with the firm offering extensive services to help companies advance their science.
💉 Evaxion Biotech received positive feedback from the International Preliminary Report on Patentability for its AI-based method identifying cancer vaccine targets using Endogenous Retroviruses and the AI-Immunology™ platform, reinforcing its position in personalized cancer vaccines.
💰 2seventy bio sells its Hemophilia A candidate and MegaTAL gene editing technology to Novo Nordisk for up to $40 million, shifting focus to its CAR T cell therapy for multiple myeloma, Abecma, in partnership with Bristol Myers Squibb.
Reinforcement Learning + LLMs = Next Gen 3D Molecular Design
Montreal-based Insilico Medicine, in collaboration with MILA—Quebec Artificial Intelligence Institute, CIFAR, and Polytechnique Montreal, introduced a new framework BindGPT leveraging reinforcement learning and LLMs to improve 3D molecular design, which uses a conceptually simple but powerful approach to create 3D molecules within the protein’s binding site.
![](https://substackcdn.com/image/fetch/w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2Fa1e15546-299b-4b48-80fc-6b0508722ed5_1630x661.png)
Key Highlights:
BindGPT generates novel 3D molecules within a protein’s binding site, merging molecular graph creation and conformation generation in one step.
BindGPT model pretrains on extensive datasets and fine-tunes with reinforcement learning, achieving results on par with the best current models at a fraction of the cost.
Versatility: BindGPT is a 3D molecular generative model, conformer generator, and pocket-conditioned molecule generator without representational equivariance assumptions.
Read the preprint: BindGPT: A Scalable Framework for 3D Molecular Design via Language Modeling and Reinforcement Learning.
How the Largest Generic Drug Manufacturer Uses AI
Teva Pharmaceuticals’s executive VP of global R&D and Chief Medical Officer Dr. Eric A. Hughes discussed the current state and practical applications of AI in the pharmaceutical industry at the 22nd BIOMED Israel Conference. He highlighted that while the industry is conservative and AI implementation is challenging, Teva already uses AI in specific areas with practical impact:
Locating Trials Subjects and Sites: AI helps identify trial sites with a history of good performance, optimizing patient recruitment and saving costs.
Diagnostics with Wearables: Wearable devices provide accurate biomechanics data for diagnosing movement disorders, potentially reducing trial sizes and costs.
Diagnosis via Mobile Phone: AI applications on mobile phones diagnose conditions like tardive dyskinesia and schizophrenia through facial movement and voice intonation analysis, aiding in patient selection for trials and ongoing monitoring.
Discovering Errors in Clinical Trials: AI improves data collection and error detection in clinical trials, identifying unusual patterns and delays.
Antibody Design without Mice: Teva collaborates with companies like Biolojic Design to create better antibodies, which are safer and more effective for human use.
Life Science E-commerce Embrases AI
Pharma and biotech R&D sector has been slow to adopt advanced e-commerce workflows compared to other consumer industries, struggling with an outdated supply chain and lack of marketplaces.
Expectedly, the advent of artificial intelligence, especially text-native models, such as large language models (LLMs), gave a boost to the development of life science e-commerce, and marketplaces.
A San Francisco-based Cromatic, a startup dedicated to facilitating outsourcing in life sciences research, has launched an innovative AI-driven tool called Sourcerer.
![](https://substackcdn.com/image/fetch/w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F2bd28fa3-0b49-4b01-800b-6b94bf3a7bbd_1003x787.png)
Sourcerer provides users with access to a comprehensive, cloud-based global database curated by Cromatic. Users can search by specific topics or input their requests for proposals (RFPs) to receive a tailored list of relevant vendors. These vendors can be contacted directly or engaged through Cromatic's platform via a 'white glove' service.
Targeted Vendor Matching
Unlike traditional search engines, Sourcerer focuses on vendors whose services align closely with users' RFPs, rather than promoting sponsored results. Vendor profiles are detailed with information on company location, size, annual revenue, and a brief description, facilitating easy comparison.
Dr. Ann Lin, cofounder and CEO of Cromatic, emphasized the tool's ability to streamline the vendor selection process, which is often complex and time-consuming. Sourcerer’s AI capabilities allow it to parse extensive RFPs, distilling project objectives and criteria to generate a precise list of necessary vendor skills. Users can then refine these criteria before conducting their search.
Cromatic, which secured $5.3 million in seed funding led by AgFunder and LifeX, offers a free basic version of Sourcerer. This version allows users to search for vendors based on a single service, such as DNA sequencing. For a more comprehensive service, users can subscribe to a monthly or annual SaaS plan, which provides significant cost savings compared to hiring consultants.
Around the same time, Scientist.com, a leading marketplace for R&D outsourcing, has launched Elisa, a large language model (LLM)-powered research assistant designed to support pharmaceutical and biotechnology research, offering instant access to marketplace information.
![](https://substackcdn.com/image/fetch/w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fsubstack-post-media.s3.amazonaws.com%2Fpublic%2Fimages%2F92996f07-e124-42e5-899a-7234075b7643_1395x1047.png)
Check out 19 Life Science Marketplaces to Try in 2024 (for paid subscribers) for a curated list of life science marketplaces for various purposes, from finding and ordering research consumables and biospecimens, to R&D outsourcing and finding talent.
Gene Editing Breakthrough: Going Beyound CRISPR
Researchers at the Arc Institute in California have identified a novel DNA-editing mechanism in bacteria, dubbed the “bridge editing” system, which promises to enable larger and more precise genomic modifications than current CRISPR-based techniques.
This new method, discovered by Dr. Patrick Hsu and his team, involves physically linking two DNA sequences, potentially allowing for extensive genome alterations.
Bridge editing could surpass the capabilities of the widely-used CRISPR-Cas9 system. While CRISPR-Cas9 has revolutionized gene editing by enabling precise targeting and mutation of specific DNA sequences, it primarily acts as a gene disruptor. Hsu’s system, on the other hand, leverages a naturally occurring mechanism in bacterial DNA, involving parasitic sequences that use a recombinase protein to link guide RNA to DNA sequences. This approach facilitates the addition, deletion, or reversal of extensive DNA segments without leaving behind unwanted DNA fragments, known as scars.
The bridge-editing system distinguishes itself by using a guide RNA that targets two specific DNA sequences instead of one, allowing for a more controlled and scarless manipulation of genomes.
This enhanced precision and control could significantly expand the scope of genomic modifications, making it feasible to undertake complex tasks such as chromosome-scale engineering in plants and animals.
Dr. Stephen Tang at Columbia University highlighted the promise of bridge editing but noted that its efficacy in human cells remains unproven. To date, the system has only been tested in bacterial cells and in vitro conditions. Nonetheless, Tang expressed optimism that with further research and adaptation, bridge editing could be made compatible with more complex organisms.
For further details, refer to the study published in Nature: DOI: 10.1038/s41586-024-07552-4.
While this research is in really early stage, I think it is a notable breakthrough for its wide-ranging disruption potential to all the CRISPR-based companies out there and the currently mainstream paradim of gene editing in general.