We would like to start from a traditional ligand-based strategy like the QSAR model and combine it with structure-based ranking. QSAR model will be developed based on historical data of known hits. Enamine Real will be screened with the QSAR model. The hit molecules will be docked, and poses will be refined with ML-accelerated QM (ML force fields) calculations and traditional MD simulations. If time permits, we will also attempt binding free energy (BFE) calculations. The data from the hit optimization step will be included in an active learning-like (AL) fashion. New data will be used to improve QSAR model and docking/BFE re-scoring. The iterative strategy will evaluated upon comparison with the results of the first round of the experiment. 

The baseline method could be straightforward large-library docking (on request from organizers) and clustering of hits.

Our group has experience and success in numerous practical drug-discovery projects. Given our experience in CACHE Challenge #1 and very different scope we suggest a modified strategy tailored for this problem. Additionally, the potential of quantum mechanics in drug discovery is underutilized due to the high cost, therefore this challenge is an excellent testbed for the realistic application. Our lab is a leading developer of ML-accelerated QM methods. 

N/A

ML: scikit-learn, xGBoost, rdkit, python, pytorch

ML&QM: python, pytorch, torchani, aimnet2, auto3d

MD: Gromacs, Amber, Amber tools

QM: ORCA

Anstine D, Zubatyuk R, Isayev O. AIMNet2: A Neural Network Potential to Meet your Neutral, Charged, Organic, and Elemental-Organic Needs. ChemRxiv. 2023; doi:10.26434/chemrxiv-2023-296ch https://chemrxiv.org/engage/chemrxiv/article-details/6525b39e8bab5d2055123f75

Gutkin E, Gusev F, Gentile F, Ban F, Koby SB, Narangoda C, et al. In silico screening of LRRK2 WDR domain inhibitors using deep docking and free energy simulations. ChemRxiv. 2023; doi:10.26434/chemrxiv-2023-lnzvr  https://chemrxiv.org/engage/chemrxiv/article-details/658485ab9138d23161354822

Scalable hybrid deep neural networks/polarizable potentials biomolecular simulations including long-range effects
Chem. Sci., 2023,14, 5438-5452
https://pubs.rsc.org/en/content/articlelanding/2023/sc/d2sc04815a

Filipp Gusev, Evgeny Gutkin, Maria G. Kurnikova, and Olexandr Isayev Journal of Chemical Information and Modeling 2023 63 (2), 583-594 DOI: 10.1021/acs.jcim.2c01052