Generative AI drug discovery hub Converge Bio raises $5.5M to understand the 'language of biology'

Drug discovery startup Converge Bio Inc. said today it has closed on a $5.5 million seed round of funding to help biotech and pharmaceutical companies develop more effective drugs at faster speeds by leveraging generative artificial intelligence.

The startup has created a novel platform that employs advanced large language models trained on biological and chemical data, such as DNA, RNA and protein sequences and other types of molecular patterns. Its LLMs leverage that knowledge to discover and develop new kinds of drugs for all kinds of diseases and medical conditions.

What Converge Bio wants to do is accelerate the drug discovery process, which has always been an extremely complex, expensive and time-consuming endeavor. As the startup points out, it takes about 10 years to develop new drugs and bring them to market, with costs often exceeding $1 billion. The expense means that a lot of promising drugs are never made, and the time it takes means that many people die while waiting for possible cures to be developed.

One of the biggest headaches in drug discovery is the failure rate, with more than 90% of candidate drugs failing to get past the clinical trial process, despite millions of dollars being spent on their development.

It's this particular problem that Converge Bio says it can solve. Co-founder and Chief Executive Dov Gertz (pictured, center, alongside Chief Science Officer Iddo Weiner and Chief Technology Officer Oded Kalev), said generative AI has incredible potential to aid in drug discovery and development, and there are already dozens of examples of new medicines created with the technology currently undergoing clinical trials.

"Biological languages, like DNA and RNA, aren't just random sequences -- they have intricate structures and rules, similar to the grammar and syntax of spoken languages," he said.

Generative AI can help researchers to understand the language of DNA and RNA, and then use it to create new medicines.

"Just like how ChatGPT can instantly generate a string of words to answer a question, a model trained on biological data can help rapidly design optimized mRNA sequences for vaccines, cutting months or years from the process," Gertz promised.

The startup has created what it says is a "generative AI hub" that biotech and pharma companies can use to accelerate the discovery and development of new, DNA- and RNA-based drugs. It offers a library of foundational LLMs for all biological languages that can be customized and fine-tuned with additional data.

Its models are designed to aid in the creation of very specific types of drugs, such as engineering antibodies with increased effectiveness and less side effects, or mRNA vaccines that are better optimized to elicit stronger immune responses. It also helps with small molecule optimization and the creation of novel proteins and new biomarkers.

Gertz said the drug discovery process starts with predicting biological outcomes, assessing how well a new medicine might work. The models provide explanations for its predictions in biological terms that scientists can understand, and this allows them to refine the drugs to create more optimized candidates that deliver better patient outcomes.

"The models are pretrained on vast amounts of data, so they can create significant insights even from minimal fine-tuning data," he said.

Bio Converge's platform has already gotten the attention of a number of pharmaceutical companies, including Teva Pharmaceutical Industries Ltd., which makes generic drugs, as well as Compugen Ltd., which specializes in immuno-oncology therapeutics, and BiomX Inc., a company that develops customized phage therapies.

TLV Partners was the only investor named in the round. Managing Partner Shahar Tzafrir said he likes to invest in startups with bold visions as a matter of course, but he was "blown away" by Converge Bio's approach.

"[It] integrates generative AI into the fabric of biology," he said. "The concept of treating biological data as a language is transformative. It's not just advancing drug discovery, but reshaping how we understand, explain and manipulate biology itself."