The unified accelerator architecture for RNA secondary structure prediction on FPGA

Abstract

In the field of RNA secondary structure prediction, MFE, SCFG and the homologous comparative sequence analysis are three kinds of classical computation analysis approaches. However, the parallel efficiency of many implementations on general-purpose computers are greatly limited by complicated data dependency and tight synchronization. Additionally, large scale parallel computers are too expensive to be used easily for many research institutes. Recently, FPGA chips provide a new approach to accelerate those algorithms by exploiting fine-grained custom design. We propose a unified parallelism schemes and logic circuit architecture for three classical algorithms—Zuker, RNAalifold and CYK, based on a systolic-like master-slave PE (Processing Element) array for fine-grained hardware implementation on FPGA. We partition tasks by columns and assign them to PEs for load balance. We exploit data reuse schemes to reduce the need to load matrix from external memory. The experimental results show a factor of 12–14× speedup over the three software versions running on a PC platform with AMD Phenom 9650 Quad CPU. The computational power of our prototype is comparable to a PC cluster consisting of 20 Intel-Xeon CPUs for RNA secondary structure prediction; however, the power consumption is only about 10% of the latter.