Enveda Biosciences Inc., a startup using artificial intelligence to develop new medicines, today announced that it has raised $55 million in funding to support its research.

Microsoft Corp. was among the more than half dozen investors that participated in the round. According to TechCrunch, the technology giant will provide Enveda with not only capital but also cloud computing credits. Microsoft participated in the round alongside Premji Invest, Lingotto Investment Fund, The Nature Conservancy and several returning Enveda backers.

It’s estimated that only a fraction of the world’s chemical compounds have been discovered to date. Boulder, Colorado-based Enveda is developing a platform that promises to help scientists “rapidly probe our planet’s dark chemical space for new medicines.” The software uses AI to speed up parts of the drug discovery process. 

The first component of Enveda’s platform is a database of small molecules, organic compounds that have a relatively low mass compared to other materials. It’s estimated that small molecules are the active ingredients in about 90% of pharmaceutical drugs. As a result, they represent a major focus area for medical researchers.

According to TechCrunch, Enveda’s database places an emphasis on providing information about molecules found in plants. So far, the company has reportedly collected data on 38,000 medicinal plants linked to about 12,000 diseases and symptoms.

According to Enveda, its researchers collect information about promising compounds using a technique known as tandem mass-spectrometry. The method involves projecting a beam of electrons onto the material being studied to turn it into ions, or charged atoms. Those ions can be studied to identify whether the material contains unknown compounds, as well as measure the quantity in which the compounds are present.

Enveda interprets the results of its tandem mass-spectrometry scans with the help of an internally-developed AI model. The model is based on the Transformer architecture, a neural network design that also underpins large language models such as GPT-4. Enveda says that its AI can automate much of the work involved in turning mass-spectrometry readings into structural formulas, the diagrams used to represent the molecular structure of chemicals.

The company’s database of small molecules is accessible through a Google-like interface. According to Enveda, scientists can find compounds with specific chemical properties by entering a natural language description of those properties. A researcher could, for example, ask the platform to surface compounds that play a role in photosynthesis and fall below a certain mass threshold.

Enveda will use the proceeds from its latest funding round to enhance its platform. Additionally, the investment will support two drug development programs through which the company hopes to discover new ways of treating skin conditions and bowel disease. Enveda plans to advance the programs to clinical trials later this year and in early 2025.

Image: Enveda