Abstract | ||
---|---|---|
Using a set of machine learning based predictors that are capable of predicting ligand-induced shielding effects on H-1 and C-13 nonexchangeable nuclei, it was discovered that holo NMR chemical shifts can be used to resolve RNA ligand poses. This was accomplished by quantitatively comparing measured and predicted holo chemical shifts in conformationally diverse "decoy" pools for three test cases and then, for each, comparing the native pose to the pose in the decoy pool that exhibited the lowest error. For three test cases, the poses in the decoy pools that exhibited the best agreement between measured and predicted holo chemical shifts were within 0.28, 1.12, and 2.38 A of the native poses. Interestingly, the predictors used in this study were trained on a database containing, only, apo RNA data. The agreement between the chemical shift-selected poses and the native NMR poses suggests that the predictors used in this study were able to "learn" general chemical shift-structure relationships from apo RNA data that could be used to account for ligand-induced shielding effects on RNA nuclei for the test cases studied. |
Year | DOI | Venue |
---|---|---|
2016 | 10.1021/acs.jcim.5b00593 | JOURNAL OF CHEMICAL INFORMATION AND MODELING |
Field | DocType | Volume |
RNA,Ligand,Decoy,Chemistry,Bioinformatics,Chemical shift | Journal | 56 |
Issue | ISSN | Citations |
2 | 1549-9596 | 0 |
PageRank | References | Authors |
0.34 | 6 | 1 |
Name | Order | Citations | PageRank |
---|---|---|---|
Aaron T. Frank | 1 | 2 | 1.71 |