2024-07-16

BIOPTIC Wins Gold at BELKA Challenge 2024

News
3

Bioptic’s own Vlad Vinogradov just brought home a gold medal at the BELKA Challenge—one of the toughest AI competitions in drug discovery. His win proves our tech and talent can go head-to-head with the best, pushing the frontier of small molecule prediction and accelerating the path to new medicines.

We’re proud to announce that Vlad Vinogradov, a key member of the Bioptic team, achieved a remarkable victory by earning a gold medal in the BELKA Challenge on July 16, 2024. Vlad’s outstanding performance placed him on the leaderboard of this prestigious competition, highlighting Bioptic’s expertise in AI-driven drug discovery. In this blog post, we celebrate Vlad’s success and explore the significance of the BELKA Challenge in advancing drug discovery.

Vlad Vinogradov’s Gold Medal Achievement

Vlad Vinogradov excelled among global participants in the BELKA Challenge, securing a gold medal and a spot on the leaderboard. His winning solution showcased his ability to predict small molecule-protein interactions with high accuracy, a critical step in drug discovery.

About the BELKA Challenge

The BELKA Challenge, officially titled “NeurIPS 2024 - Predict New Medicines with BELKA,” was a Kaggle competition hosted by Leash Biosciences, a discovery-stage biotech company based in Salt Lake City, Utah. The challenge tasked participants with developing machine learning models to predict the binding affinity of small molecules to three protein targets: BRD4, EPHX2/sEH, and ALB/HSA. These predictions are vital for drug discovery, as small molecule drugs often work by inhibiting specific protein targets involved in disease processes.

Leash Biosciences provided the Big Encoded Library for Chemical Assessment (BELKA) dataset, which includes 133 million small molecules and approximately 100 million binding measurements, making it one of the largest datasets of its kind. Unlike the smaller BindingDB dataset (2.8 million measurements), BELKA offers a rare opportunity to explore the vast drug-like chemical space—estimated at 10^60 possible compounds—using AI rather than traditional, labor-intensive lab testing. Participants could use any method, from data-driven ML models to simulations like DiffDock, to predict binding affinities.

An official NeurIPS 2024 competition, the BELKA Challenge aimed to democratize computational drug discovery by providing access to a dataset typically exclusive to large pharmaceutical companies. The results were presented at the NeurIPS 2024 conference in Vancouver, Canada, highlighting the winning teams’ contributions to advancing small molecule chemistry and accelerating the discovery of new medicines.

Impact on Bioptic and the Industry

Vlad’s success in the BELKA Challenge reinforces Bioptic’s commitment to advancing AI-driven drug discovery. His expertise aligns with our mission to explore vast chemical spaces, as demonstrated by our work on selective kinase inhibitors and collaborations like the recent $17.5 million deal with Vindur Tx. The BELKA dataset and competition have set a new standard for the industry, encouraging the development of AI models that can navigate immense chemical spaces to find viable therapeutic candidates more efficiently.

Get in touch

Whether you’re a researcher, potential partner, or just curious about what we’re building, drop us a message. Explore how we can push the boundaries of science and discovery.