FPGA implementation of high‐speed tunable IIR band pass notch filter for identification of hot‐spots in protein

• Published in: International journal of circuit theory and applications 2021-11, Vol.49 (11), p.3748-3765
• Main Authors: Pathak, Vikas, Nanda, Satyasai Jagannath, Joshi, Amit Mahesh, Sahu, Sitanshu Sekhar
• Format: Article
• Language: English
• Subjects: Amino acids; band pas notch IIR digital filter; Digital filters; Digital signal processing; Field programmable gate arrays; floating point number; Fourier transforms; FPGA implementation; Functional groups; Hardware; hot‐spot locations in proteins; IIR filters; Impulse response; Matlab; Notch filters; Proteins; proteomics; Sequences
• ISSN: 0098-9886; 1097-007X
• DOI: 10.1002/cta.3131

Summary Localization of hot spots in a protein sequence is an important problem for understanding complex 3D structure of amino acid sequence (due to which biological functionality of protein is performed by their mutual interactions). Several digital signal processing (DSP) techniques like digital filtering, wavelet, and Fourier transforms have been used for finding the hot‐spot locations. Fast processing of these longer protein sequences is desirable. Field Programmable Gate Array (FPGA) implementation of Infinite Impulse Response (IIR) band pass notch (BPN) digital filter (tuned automatically by characteristics frequency of protein family) is carried out in this paper for further speed‐up of this protein analysis process. Performance of proposed filter system (implemented on Zynq series Zybo board FPGA) is further improved using retiming, which has significantly increased the maximum clock frequency from 34.304 to 54.812 MHz. Simulation of both MATLAB and FPGA hardware implementation for BPN IIR filter is executed on 10 protein functional groups. The simulation results predict the same hot‐spots locations as mentioned in ASEdb data base, but the Hardware BPN IIR filter (proposed in this paper) is 190 to 278 times faster as compared to its MATLAB counterpart. Some additional new hot‐spot locations (not mentioned in ASEdb data base) have been determined by our proposed hardware filter, which can be the probable hot‐spots for newly identified proteins. Hot‐spots localization in protein sequence is important for researchers, as biological functionality of protein is performed by their mutual interactions at these hot‐spots. In this manuscript, FPGA implementation of band pass notch IIR digital filter is carried out, which is used in high‐speed identification of hot spots in proteins. A retiming technique is used for performance improvement of the proposed hardware architecture. The proposed design is 190 to 278 times faster for hot‐spots identification in various proteins over its MATLAB implementation.