ABSTRACT
Abstract Contrary to heavy genome as well as proteome sequencing rates, the pace of experimental solving of protein structures is quite low and the sequence-structure gap is constantly increasing. Detailed structural knowledge of a protein is essential to understand its native function in a cell and it consequently helps us to learn the expression profile of all the genes in that cell. Numerous computational algorithms for protein structure prediction have been developed to quickly construct the protein models and to bridge this ever-increasing sequence-structure gap. However, the current prediction methodologies often fail to select the true and acceptable conformation from the generated decoy structures. Also, the currently popular model ranking schemes are not efficacious in resolving very close structural models to identify the actual model even when the experimental structure of the considered sequence is available to aid in the decision making. This chapter extensively investigates the current best model assessment measures, including those which evaluate structure and geometry, such as GDT_TS, GDT_HA, Spheregrinder, TM-Score and RMSD scores, and illustrates their inaccuracies in effectively resolving and ranking the protein models. It further presents a new method of ranking the constructed protein models on the basis of topological differences in the dihedral angles and the distances between successive Ca atoms. The developed methodology is inspected by employing it for re-ranking the top five models for five TBM-Easy targets and five TBM-Hard targets from the CASP10 database, and it proves the inefficiency of the currently used model assessment measures in selecting the accurate structure for a target protein sequence. The ranking result is further verified by evaluating the atomic clashes in these models. The resultant model ranking is intelligently utilized to apprehend the coherent shortcomings of contemporary assessment measures and expound the improved model assessment-cum-ranking algorithm. Furtherance of model evaluation and ranking measures will certainly help us to consistently select the best predicted conformations for understanding a protein’s functional role in the biological pathways of a cell system.
