TY - GEN
T1 - Accu RMSD
T2 - 5th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics, ACM BCB 2014
AU - Akbal-Delibas, Bahar
AU - Pomplun, Marc
AU - Haspel, Nurit
N1 - Publisher Copyright:
Copyright © 2014 ACM.
PY - 2014/9/20
Y1 - 2014/9/20
N2 - Protein-protein docking methods aim to compute the correct bound form of two or more proteins. One of the major challenges for docking methods is to accurately discrimi- nate native-like structures. The protein docking community agrees on the existence of a relationship between various favorable intermolecular interactions (e.g. Van der Waals, electrostatic, desolvation forces, etc.) and the similarity of a conformation to its native structure. Different docking algo- rithms often formulate this relationship as a weighted sum of selected terms and calibrate their weights against a specific training data to evaluate and rank candidate structures. However, the exact form of this relationship is unknown and the accuracy of such methods is impaired by the pervasiveness of false positives. Unlike the conventional scoring functions, we propose a novel machine learning approach that not only ranks the candidate structures relative to each other but also indicates how similar each candidate is to the native conformation. We trained the AccuRMSD neural network with an exten- sive dataset using the back-propagation learning algorithm and achieved RMSD prediction accuracy with less than 1fiA error margin on 19,600 test samples.
AB - Protein-protein docking methods aim to compute the correct bound form of two or more proteins. One of the major challenges for docking methods is to accurately discrimi- nate native-like structures. The protein docking community agrees on the existence of a relationship between various favorable intermolecular interactions (e.g. Van der Waals, electrostatic, desolvation forces, etc.) and the similarity of a conformation to its native structure. Different docking algo- rithms often formulate this relationship as a weighted sum of selected terms and calibrate their weights against a specific training data to evaluate and rank candidate structures. However, the exact form of this relationship is unknown and the accuracy of such methods is impaired by the pervasiveness of false positives. Unlike the conventional scoring functions, we propose a novel machine learning approach that not only ranks the candidate structures relative to each other but also indicates how similar each candidate is to the native conformation. We trained the AccuRMSD neural network with an exten- sive dataset using the back-propagation learning algorithm and achieved RMSD prediction accuracy with less than 1fiA error margin on 19,600 test samples.
KW - Machine learning
KW - Neural networks
KW - Protein docking and refinement
KW - Rmsd prediction
KW - Scoring functions
UR - https://www.scopus.com/pages/publications/84920747678
UR - https://www.scopus.com/pages/publications/84920747678#tab=citedBy
U2 - 10.1145/2649387.2649392
DO - 10.1145/2649387.2649392
M3 - Conference contribution
AN - SCOPUS:84920747678
T3 - ACM BCB 2014 - 5th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics
SP - 289
EP - 296
BT - ACM BCB 2014 - 5th ACM Conference on Bioinformatics, Computational Biology, and Health Informatics
PB - Association for Computing Machinery
Y2 - 20 September 2014 through 23 September 2014
ER -