Basecamp Research, a biotechnology company that uses artificial intelligence to discover novel proteins and to help create new medicines, said today it has completed a $60 million round and joined a multiyear collaboration with the Broad Institute of Massachusetts Institute of Technology and Harvard.
The company’s Series B financing round was led by Singular, with additional investors joining including S32, redalpine, True Ventures and Hummingbird Ventures, vice-chairman of Roche André Hoffmann, chair of Royal Philips Feike Sijbesma and Paul Polman, former CEO of Unilever.
Basecamp has been working to build an extensive dataset of proteins found in nature, with a vision of creating a knowledge graph that will feed AI algorithms to promote the discovery of unexplored drugs and materials. It does this using what the company calls a “foundational database tailored to AI.”
“We are pushing past AI’s limits in biological design by doubling down on solving the fundamental data gap that the entire biotech industry faces,” said Dr. Glen Gowers, co-founder and chief executive of Basecamp Research. “Basecamp Research is using its technology to generate new and deeper insights, going beyond known biology and each day expand what we can offer to our partners in the biopharma ecosystem.”
By joining the Broad Institute, Basecamp is collaborating with Dr. David R. Liu, a Howard Hughes medical institute investigator and core member at Broad, and the Liu Lab to create new approaches to “programmable genetic medicines.” These medicines could include therapies that interact with biological systems to regulate doses, change their method of action or be designs based on specifically coded proteins from the genes of a population or disease. The objective is to create treatments for diseases by developing new methods of discovering fusion proteins and other large molecules to produce the next generation of drugs.
Basecamp’s transformative work has already led to the creation of new AI models for the discovery of new protein molecules using its technology. According to the company, although large language models such as ChatGPT have proven useful for helping scientists design and work with protein sequences, they require significant training and conditioning on known protein starter sequences.
To jumpstart this process, the company partnered with Ferruz Laboratory at the Institute of Molecular Biology of Barcelona and announced the release of ZymCtrl in June. The company said it’s a next-generation end-to-end protein LLM that provides design capabilities for generating artificial enzymes, a protein that speeds up chemical reactions in living cells. Unlike other LLMs, it requires no seed sequence and it can create enzymes that work and share only 30% resemblance to its training set.
“With ZymCtrl, generating highly specific enzymes is as easy as interacting with a chatbot,” said Noelia Ferruz, group leader of the laboratory, who has been partnering with Basecamp for more than two years.
The company also launched BaseFold, a deep learning model designed to predict the 3D structures of large, complex proteins more accurately than other models. The company said it created the new model by augmenting AlphaFold2, the industry gold standard for AI-powered protein prediction. BaseFold has proven six times more accurate than AlphaFold2 and provides a three-fold improvement in small molecule docking, the company added.
Basecamp said it will use the proceeds from the fundraise to scale up its data collection efforts by growing its foundational dataset, which already contains 100 times more advanced biological systems information than public databases. The company also intends to strengthen its research into AI capabilities and focus on a new generation of foundational models for protein discovery.