IN-SILICO PREDICTION OF PROTEIN-PEPTIDE INTERACTIONS Automated docking methods are used extensively for gaining a mechanistic understanding of the molecular interactions underpinning cellular processes. While these tools work well for small molecules they perform poorly for peptides and cannot handle Intrinsically Disordered Proteins (IDPs) which play very important roles in these processes. The goal of this project is the development of an efficient and practical peptide docking software, useful for designing therapeutic peptides and gaining insight into IDPs binding ordered proteins. The proposed software supports biomedical applications ranging from investigating chemical pathways to designing and optimizing therapeutic molecules for diseases such as cancer and metabolic disorders. Under the previous award we developed and released a new method for docking fully-flexible peptides with up to 20 standard amino acids: AutoDock CrankPep (ADCP). We showed that it outperforms current state-of-the-art docking methods. For the next award, we propose to: 1) further develop ADCP to support docking IPDs with up to 70 amino acids and improve support for therapeutic peptides containing modified amino acids and complex macrocycles; 2) develop peptide-specific scoring functions to increase docking success rates and methods for predicting the free energy of binding of peptides. This will be done by exploiting the latest advances in statistical potentials for docking, as well as applying machine-learning techniques; 3) test and validate the software on our datasets, community benchmarks, and through our collaborations with outstanding biologists working on biomedical applications spanning from designing drugs for thrombosis and influenza, to modeling IDPs interacting with globular proteins; and 4) document the software and release it under an open source license on a regular basis along with datasets we compile and update on regularly. The proposed research will occur in the context of collaborations with experimental biologists working on highly relevant biomedical projects and providing experimental feedback and validation. In addition, this project will benefit from various collaborations with experts in the fields of computational biology, applied mathematics and artificial intelligence. This docking software tool will be developed by applying best practices in software engineering and be implemented as a modular, extensible, component-based software framework for peptide docking. This docking engine will be part of the widely used AutoDock software suite. The ability to model complexes formed by proteins and fully-flexible peptides or IDPs is in high demand and will greatly extend the range of peptide-based therapeutic approaches for which automated docking can be successfully applied. It will also support gaining insights into interactions of IDPs with proteins. As such, it will impact the research of many medicinal chemists and biologist and extend the use of computational tools to a wider community of scientists, thereby supporting the advancement of biomedical research.
Automated docking is a workhorse for rational drug design, however, applying these methods to peptides has remained challenging, thus impeding the designing of therapeutic peptides and the study of Intrinsically Disordered Proteins (IDP) binding to their ordered partners. During the prior funding period, we made substantial progress toward peptide docking, resulting in a new docking engine: AutoDock CrankPep, which outperforms state-of-the-art docking methods for linear and cyclic peptides with up to 20 standard amino acids. We propose to further develop AutoDock CrankPep to support docking of therapeutic peptides with modified amino acids as well as IDPs with up to 70 amino acids, creating a practical docking tool for peptides that will impact the research of many computational and medicinal chemists and biologist, contribute to our understanding of biological processes, and significantly advance biomedical research.
| Fernández, José A; Xu, Xiao; Sinha, Ranjeet K et al. (2017) Activated protein C light chain provides an extended binding surface for its anticoagulant cofactor, protein S. Blood Adv 1:1423-1426 |
| Ravindranath, Pradeep Anand; Sanner, Michel F (2016) AutoSite: an automated approach for pseudo-ligands prediction-from ligand-binding sites identification to predicting key ligand atoms. Bioinformatics 32:3142-3149 |
| Forli, Stefano; Huey, Ruth; Pique, Michael E et al. (2016) Computational protein-ligand docking and virtual drug screening with the AutoDock suite. Nat Protoc 11:905-19 |
| Ravindranath, Pradeep Anand; Forli, Stefano; Goodsell, David S et al. (2015) AutoDockFR: Advances in Protein-Ligand Docking with Explicitly Specified Binding Site Flexibility. PLoS Comput Biol 11:e1004586 |
