Dramatic advances in our understanding of molecular structure and function promise to accelerate the creation of new diagnostics and therapeutics. However the link between the structure of a biological macromolecule and its function is usually not obvious: fundamental to understanding how a molecule functions is an understanding of how its structure behaves over time. Recent advances in molecular dynamics simulations now allow the rapid collection of information about structural motion. These data sets are huge, and require statistical machine learning algorithms to characterize and recognize patterns relevant to function. The National Library of Medicine's new long-range plan calls for research in the use of advanced simulation and machine learning algorithms in support of biomedical research. This proposal focuses on annotating molecular structures with missing or incomplete functional information. We are particularly interested in identifying binding sites and active sites in proteins. We bring together simulation and machine learning, and hypothesize that the performance of structure- based function annotation methods will dramatically improve with the addition of information about dynamics. Thus, our specific aims are (1) to develop methods for recognizing function from structural dynamics and diversity, (2) to develop capabilities for large scale clustering and analysis tools for the discovery of novel functions, and (3) to apply our tools to challenging and important biological systems, while disseminating our software, data and capabilities to the biomedical research community. In particular, we will focus our new capabilities on three difficult function annotation challenges: ATP binding sites, phosphorylation sites, and metabolizing enzyme active sites.
The explosion in data related to molecular biology has created great opportunities for new disease diagnostics and therapies. One source of data is the three-dimensional (3D) structure of biological molecules such as proteins, DNA and RNA. This work focuses on using computational technologies to understand how these structures perform their function, so we have a better understanding of both normal and disease processes.
| Petkovic, Dragutin; Altman, Russ; Wong, Mike et al. (2018) Improving the explainability of Random Forest classifier - user centered approach. Pac Symp Biocomput 23:204-215 |
| Mallory, Emily K; Acharya, Ambika; Rensi, Stefano E et al. (2018) Chemical reaction vector embeddings: towards predicting drug metabolism in the human gut microbiome. Pac Symp Biocomput 23:56-67 |
| Zhou, Weizhuang; Altman, Russ B (2018) Data-driven human transcriptomic modules determined by independent component analysis. BMC Bioinformatics 19:327 |
| Lo, Yu-Chen; Rensi, Stefano E; Torng, Wen et al. (2018) Machine learning in chemoinformatics and drug discovery. Drug Discov Today 23:1538-1546 |
| Previde, Paul; Thomas, Brook; Wong, Mike et al. (2018) GeneDive: A gene interaction search and visualization tool to facilitate precision medicine. Pac Symp Biocomput 23:590-601 |
| Liu, Tianyun; Ish-Shalom, Shirbi; Torng, Wen et al. (2018) Biological and functional relevance of CASP predictions. Proteins 86 Suppl 1:374-386 |
| Percha, Bethany; Altman, Russ B (2018) A global network of biomedical relationships derived from text. Bioinformatics 34:2614-2624 |
| Lavertu, Adam; McInnes, Greg; Daneshjou, Roxana et al. (2018) Pharmacogenomics and big genomic data: from lab to clinic and back again. Hum Mol Genet 27:R72-R78 |
| Gottlieb, Assaf; Daneshjou, Roxana; DeGorter, Marianne et al. (2017) Cohort-specific imputation of gene expression improves prediction of warfarin dose for African Americans. Genome Med 9:98 |
| Torng, Wen; Altman, Russ B (2017) 3D deep convolutional neural networks for amino acid environment similarity analysis. BMC Bioinformatics 18:302 |
Showing the most recent 10 out of 64 publications
