Identifying potential medicinal compounds from the vast number of molecules that exist presents a significant challenge, but biotechnology companies are adopting a more efficient strategy: working with greater intelligence, not increased effort. Genesis Therapeutics has developed a novel simulation methodology and assembled a highly skilled, interdisciplinary team, achieving notable success with a recently secured $52 million Series A funding round.

Last year, Genesis showcased its potential at the Startup Battlefield during Disrupt, captivating the judges and attracting substantial interest, notably from Rock Springs Capital, who spearheaded the investment round.

In recent years, numerous companies have emerged within the drug discovery sector, fueled by advancements in computational power and simulation capabilities that allow for the assessment of molecules' therapeutic potential for various diseases. While this is the underlying principle, the practical application is more intricate. These companies can refine the search process, but they cannot definitively identify, for example, a cure for Parkinson’s disease.

The company’s founder, Evan Feinberg, entered the field after a personal health challenge—an inherited condition—made traditional laboratory work difficult during an internship at a major pharmaceutical firm. He found the computational aspects of the field more accessible and ultimately dedicated himself to it.

Prior experience led him to a significant advancement in the digital representation of molecules. While machine learning has accelerated progress in numerous fields, including biochemistry, he believed the technology’s full capabilities remained untapped.

“Early attempts often involved adapting deep learning techniques and representing molecules similarly to images, for classification purposes—determining whether an image contains a cat, for instance,” he explained in an interview. “Our approach represents molecules in a more natural way: as graphs. These graphs consist of nodes, representing atoms, and connections between them, representing bonds. However, we go beyond simply identifying the presence or absence of a bond, incorporating multiple contact types between atoms, spatial distances, and other complex characteristics.”

This resulting representation is more detailed and comprehensive, providing a more complete molecular profile than a chemical formula or a simple structural diagram. Biochemical processes are rarely as straightforward as a diagram suggests. Each molecule exists as a dynamic, three-dimensional structure, where critical factors like the distance between carbon formations or bonding sites are influenced by numerous variables. Genesis strives to model as many of these factors as possible.

“The initial step is the representation,” he stated, “but the logical next step is to utilize that representation to develop a function that takes an input and produces an output, such as binding affinity or solubility, or a vector that predicts multiple properties simultaneously.”

The company’s primary focus is on this endeavor—not merely creating a superior molecular model, but also being able to simulate a theoretical molecule and predict its behavior, identifying its strengths and limitations.

Some of this work will be conducted through collaborations, such as the partnership with Genentech, but the team is also establishing an independent internal development process to potentially discover drug candidates independently.

The $52 million in funding will significantly contribute to these efforts, as Feinberg noted in an email:

“These funds will enable us to pursue several key objectives, most importantly continuing to pioneer artificial intelligence technologies for drug development and advancing our therapeutic pipeline. We will be expanding our team by hiring leading AI researchers, software engineers, medicinal chemists, and biotechnology professionals, and establishing our own research laboratories.”

While other companies are pursuing similar simulation-based approaches, Feinberg believes Genesis possesses at least two distinct advantages, despite the competition having raised substantial funding and operating for a longer period.

“We are the only company in this field integrating modern deep neural network methodologies with biophysical simulation—specifically, the conformational changes of ligands and proteins,” he said. “Furthermore, we are applying this highly technical platform to the expertise of professionals who have successfully brought FDA-approved drugs to market. The synergy between these areas has already yielded significant value, with chemists contributing to the development of the AI as well.”

The recent advancements in AlphaFold, which rapidly simulates protein folding with unprecedented speed, are as encouraging to Feinberg as they are to the broader scientific community.

“As scientists, we are very enthusiastic about the recent progress in protein structure prediction. It represents an important advancement in basic science that will ultimately have significant downstream benefits for the development of new therapeutics,” he wrote. “Since our Dynamic PotentialNet technology uniquely leverages 3D structural information of proteins, computational protein folding—similar to recent progress in cryo-EM—is a beneficial complement to the Genesis AI Platform. We commend all efforts to make protein structure more accessible, facilitating the development of therapeutics for patients with all conditions.”

T. Rowe Price Associates, Andreessen Horowitz (who led the seed round), Menlo Ventures and Radical Ventures also participated in the funding round.