MSD's $345m alliance with Variational, and other AI news

MSD has signed an agreement with Variational AI to use its generative AI (genAI) platform Enki in the discovery of new small-molecule medicines.

Vancouver, Canada-based Variational – which emerged in 2021 with seed funding of $3.4 million – will use a "fine-tuned" version of Enki to find candidates against two undisclosed targets in the deal, which could be worth as much as $349 million in upfront and milestone payments.

According to the company, Enki has been purpose-built as a drug discovery-focused genAI platform, trained on molecular structures and properties from hundreds of millions of curated experimental and computational samples from more than 700 drug targets, and does not rely on known structures or compound libraries.

It focuses on activities like lead optimisation carried out prior to preclinical development and, according to a benchmarking study reported last year, can generate 100 hits within weeks with a value as effective as a conventional high-throughput screen of more than a million molecules.

MSD (known as Merck & Co in the US and Canada) joins smaller biopharma companies using the platform, including ImmVue, Rakovina Therapeutics, and OncoCross.

The pharma group's head of discovery chemistry, Robert Garbaccio, said the alliance is part of MSD's drive "to harness the potential of AI to improve efficiency, speed, and quality of candidates earlier in the discovery continuum," adding the company will work with Variational on "challenging therapeutic targets."

Also this week

In other recent news on the application of AI to biopharma discovery, Swiss 'decentralised science' (DeSci) specialist Bio Protocol raised $6.9 million in funding for the development of its BioAgents, described as "autonomous AI co-scientists" that can generate hypotheses and simultaneously bring in funding for projects by marrying AI, biotech, and crypto.

A blog post from the company said the funding will "fuel […] growth into a full-stack platform for AI-driven DeSci, giving researchers, patients, and the crypto community new ways to build and monetise biotech discoveries outside traditional pharma structures and faster than legacy biotech."

A new UK start-up called Dalton Tx emerged from stealth with seed financing of £4 million ($5.4 million) and a mission to build adaptive AI platforms for the full discovery process, from "raw data and model training to molecular design, synthesis, and decision-making."

The London company, based at the King's Cross life sciences and tech hub, is led by former scientists at AstraZeneca, Exscientia, and the University of Oxford, and says its platform is designed to support both small-molecule and biologic programmes.

Scientists at Virginia Tech in the US have developed an AI tool, ProRNA3D-single, that can predict and visualise what's going on at the RNA level inside patients infected by viruses or succumbing to neurological diseases like Alzheimer's disease.

Described as a "bilingual ChatGPT for biology", the large language model (LLM) was derived from two others designed for proteins and RNA sequences, and works by letting them "talk" to each other. The team hopes to continue development of the tool to improve its accuracy and get even more detailed models of various biological processes.

Finally, Insilico Medicine will use its Pharma.AI drug discovery platform to develop a library of antibody-drug conjugates (ADCs) in partnership with drug developer Mabwell Bioscience and contract development and manufacturing organisation (CDMO) ChemExpress.

The aim is to pioneer a new model of ADC development, which enables the prediction of drug potency and safety at the molecular design stage, shortening the time required for molecular optimisation, a process that can take years.