Unsolved problems of ambient computationally intelligent TBM algorithms

Abstract

Structural and functional characterization of protein sequences is one of the important areas of biological research. Currently, a small number of experimentally solved protein structures exist in Protein Data Bank (PDB) in comparison to their considerably higher count of sequence available in UniProtKB/Swiss-Prot. Ambient template-based modelling (TBM) algorithms computationally predict conformational details of a protein sequence on the basis of its evolutionary related similarity with other experimentally solved protein structures. Despite several improvements, shortcomings still obstruct the accuracy of every single step of a TBM algorithm. In this study, we discuss the shortcomings as well as probable corrective measures of major TBM algorithm steps like search and selection of the reliable templates, construction of an accurate target-template alignment, model building, and sampling, and model assessment for selecting the best conformation.