AI finds its place in all areas of biotechnology and pharmaceutical research, but as in other industries, its implementation is never as simple as we would like. Converge Organic has built a tool that allows companies to make their biology-focused LLMs actually work, from “enriching” their data to explaining their answers. The company raised $5.5 million in a funding round to scale its product.

“A model is just a model. It’s not enough,” said CEO and co-founder Dov Gertz. “We need to create a pipeline so that companies can actually use the model in their own R&D process. The market is very fragmented, but the pharmaceutical industry and biotechnology want to consume this technology in a consolidated way, in one place. We want to be that place.

If you’re not a machine learning engineer working in drug discovery, you may not be familiar with this problem. But fundamentally, there are powerful fundamental models, large language models trained not on books and the Internet, but on huge databases of DNA, protein structures and genomics.

These are powerful and versatile models, but like the LLMs used in products like ChatGPT and Cursor, they require a lot of work to get into a form that people can actually use on a daily basis. This work is particularly difficult in specialized fields such as microbiology or immunology. Taking a “raw” LLM trained on billions of protein sequences and turning it into something that a lab technician can use as part of their normal research is a non-trivial problem.

As an example, Gertz suggested antibody research. An LLM trained in specific antibody biology exists, but it is very general. Converge Bio offers a series of enhancements that can be done securely and using a company’s own IP address.

The first is “data enrichment,” augmenting the LLM of antibodies with important related data such as antigen-antibody and protein-protein interactions. Then, loaded with more specific knowledge, it can be refined to the specific antigen the team is looking to target and may have proprietary data on in the box.

“Now we have an application: the input is a sequence, the output is a binding affinity,” Gertz said. Next, the platform provides another important layer: explainability. Researchers can explore the results to discover not only that “this sequence works better than this one” but also pinpoint down to the amino acid or base pair level what part of the sequence appears to be. manufacturing it works better.

Finally, it generates new sequences that provide improved, also explainable results. Gertz noted that Explainability has surprised them with its popularity with customers – which makes sense, because it allows experts to apply their domain expertise (e.g., protein interactions) to this newer, more advanced region. obscure bioinformatics and machine learning.

Converge uses the many open source and free foundation templates, but is also working to create its own. It already has a proprietary process, Gertz said, for the explainability part. And the data enrichment “program” is entirely up to them too – it’s not a trivial process. Training methodologies, he pointed out, are one of the few jealously guarded secrets of the most successful AI companies.

This is part of the moat they hope to build, in addition to the fact that. As Gertz says: “This is probably the biggest opportunity in biotechnology in five decades. »

Yet many, perhaps most, biotech companies do not have a dedicated solution for performing LLM-related work in their field and actively pursuing niches to which generalist solutions do not apply.

“The idea is to be the everything store for genAI in biotech, and then use that as a corner to offer more over time,” Gertz said. “The behavior in the pharmaceutical and bio sectors is that once they have a connection with a supplier they trust, they want to use it in other use cases, whether antibody design or vaccine design. This is why I think this positioning is the best on the market at the moment.

Investors appear to agree, investing $5.5 million in a seed round led by TLV partners.

The company will use the money to hire and acquire customers, as startups often do at this stage, but will also publish a scientific paper on antibody design (using its own systems, of course) and form “a appropriate basic model”. »