CVC Seminar Wed Nov 9 1:15-2:30pm ACES 4.304
Speakers: Muhibur Rasheed and Qiming Yuan
Title: Training F2Dock
F2Dock is a rigid body docking software which has a FFT based on-the-fly scoring scheme involving shape complementarity, electrostatics, simple charge complementarity and interface propensity, followed by a rescoring and reranking phase using proximity clustering, clash filter, Lennard-Jones filter, antibody filter, enzyme filter, residue contact filter, interface propensity filter, interface area filter etc. Overall there are more than 100 parameters in the scoring function. In the first part of the talk, we discuss all of these and explain the (mostly manual) strategy applied to train them. In the second part of the talk we explore several existing techniques for automated training of different types of parameters and evaluate their feasibility in training the parameters of F2Dock.