Join us at the Fragment-Based Lead Discovery conference from 21-24 Sept 2025 in Cambridge, UK. Evotec experts will be presenting their latest research and innovations. We look forward to engaging with the FBLD community. Get in touch to schedule a meeting!
Research Presented at FBLD
Posters
Machine Learning-Supported Fragment Hit Expansion in Absence of X-Ray Structures
Date and Time: Monday 22nd September, 2025; 6:10 PM-7:30 PM
Location: Fitzwilliam College at Cambridge University, UK - Reddaway and Upper Hall 1 & 2 Rooms
Presented by: Florian Krieger PhD, VP Biophysics
Fragment-based screening is a widely used method in early drug discovery, but its reliance on high-resolution X-ray structures for labour-intensive follow-up poses a challenge. To address this, Evotec has developed a combined machine learning strategy that integrates predictions from two independently trained models. One model leverages bioactivity-derived HTS fingerprints from 450 historical HTS campaigns, while the other uses structural fingerprints. These models were applied to a dataset comprising of 20,000 data points previously recorded with surface plasmon resonance (SPR), and were used to identify promising compounds from a 400,000 lead-like library. The result was a selection of 1,700 compounds which exhibited an improved rate of primary hits and specific binders compared to previously recorded fragment screening data. This approach opens up new avenues to efficiently expand weak fragment hits into the lead like chemical space for protein targets which are not amenable for X-ray crystallography.
Presentations
GCI and ML to Rapidly Transport Chemistry from Fragments to Lead-like Compounds for SLCs
Date and Time: Tuesday 23rd September, 2025; 11:30 AM-12:00 PM
Location: Fitzwilliam College at Cambridge University, UK - Auditorium
Presented by: Alicia Churchill-Angus PhD, Senior Scientist, Biophysics
Solute carrier (SLC) transporters are highly relevant in drug discovery due to their vital role in many biological processes. However, challenges in assay development and protein production, limit effective drug discovery on this target class of proteins. To address this, Evotec developed a novel workflow for SLC characterization establishing a successful protein production route to generate high quality materials. In addition, a real-time kinetics assay was developed using Grated Coupled Interferometry (GCI), and a 3,000-fragment library was screened using this approach. Validated hits informed a machine learning (ML) program that was used to select 1,000 lead-like compounds from a 250,000-compound library. This ML-guided approach achieved a 4× higher hit rate when compared to random screening, leading to the first lead-like binders for this SLC target. The strategy offers a powerful solution to challenging protein targets when conventional hit finding methods could not be established.
