Wojtek Treyde

Wojtek Treyde

he/him

PhD Student in Computational Chemistry

Publications

Journal Articles

Bond dissociation energies of X–H bonds in proteins

Published in RSC Advances, 2022

Knowledge of reliable X–H bond dissociation energies (X = C, N, O, S) for amino acids in proteins is key for studying the radical chemistry of proteins. X–H bond dissociation energies of model dipeptides were computed using the isodesmic reaction method at the BMK/6-31+G(2df,p) and G4(MP2)-6X levels of theory. The density functional theory values agree well with the composite-level calculations. By this high level of theory, combined with a careful choice of reference compounds and peptide model systems, our work provides a highly valuable data set of bond dissociation energies with unprecedented accuracy and comprehensiveness. It will likely prove useful to predict protein biochemistry involving radicals, e.g., by machine learning.

Recommended citation: W. Treyde, K. Riedmiller, F. Gräter, "Bond dissociation energies of X–H bonds in proteins." RSC Advances, 2022, 12, 34557-34564.
Download Paper

Conference Papers

QuickBind: A Light-Weight And Interpretable Molecular Docking Model

Published in Proceedings of the 19th Machine Learning in Computational Biology meeting, 2024

Predicting a ligand’s bound pose to a target protein is a key component of early-stage computational drug discovery. Recent developments in machine learning methods have focused on improving pose quality at the cost of model runtime. For high-throughput virtual screening applications, this exposes a capability gap that can be filled by moderately accurate but fast pose prediction. To this end, we developed QUICKBIND, a light-weight pose prediction algorithm. We assess QUICKBIND on widely used benchmarks and find that it provides an attractive trade-off between model accuracy and runtime. To facilitate virtual screening applications, we augment QUICKBIND with a binding affinity module and demonstrate its capabilities for multiple clinically-relevant drug targets. Finally, we investigate the mechanistic basis by which QUICKBIND makes predictions and find that it has learned key physicochemical properties of molecular docking, providing new insights into how machine learning models generate protein-ligand poses. By virtue of its simplicity, QUICKBIND can serve as both an effective virtual screening tool and a minimal test bed for exploring new model architectures and innovations. Model code and weights are available at this GitHub repository.

Recommended citation: W. Treyde, S. C. Kim, N. Bouatta, M. AlQuraishi, "QuickBind: A Light-Weight And Interpretable Molecular Docking Model." Proceedings of the 19th Machine Learning in Computational Biology meeting, PMLR, 2024, 261, 129-152.
Download Paper