FPGA-based hardware accelerator for the prediction of protein secondary class via fuzzy K-nearest neighbors with Lempel–Ziv complexity based distance measure

Abstract

Correct prediction of protein secondary structural classes is vital for the prediction of tertiary structures and understanding of their function. Most of the prediction algorithms require lengthy computation time. Nearest neighbor – complexity distance measure (NN-CDM) algorithm was one of the significant prediction algorithms using Lempel–Ziv (LZ) complexity-based distance measure, but it is slow and ineffective in handling uncertainties. To solve the problems, we propose fuzzy NN-CDM (FKNN-CDM) algorithm that incorporates the confidence level of prediction results and enhance the prediction process by designing hardware architecture that implements the proposed algorithm in an FPGA board. Highest average prediction accuracies for Z277 and 25PDB datasets using proposed algorithm are 84.12% and 47.81% respectively, with 15 times faster computation using an Altera DE2-115 FPGA board. • Fuzzy KNN with Complexity-based Distance Measure algorithm is proposed. • We model the algorithm using FPGA device on Z277 and 25PDB datasets. • Parallel processing capability of FPGA is used to improve the prediction speed. • Approximately 15 times faster computation speed can be achieved. • The algorithm is more effective for high sequence homology protein dataset.