Implementation of a dna compression algorithm using dataflow computing

Abstract

RESUMEN: The amount of DNA sequences databases has increased a lot in the last years, the amount of space required to store the sequences is increasing more than the space available to store them, that means a higher cost to store DNA sequences and also the read sequences which are fragments of the whole sequence. This situation has led to the use of compression algorithms for storing DNA files. The main objective of the project is to increase the efficiency of the compression of DNA sequences because the process requires a lot of compute. An FPGA with dataflow architecture has been used to develop the project with the aim of exploiting the available parallelism in the algorithm chosen. The compression method has been developed to process sequence reads with a fixed amount of mutations per read and the test has been developed for 4, 8, 12 and 16 mutations per reads using an architecture that allows up to 160 reads to be processed in only one thick. Experimental results showed that even with a low amount of processing units, the performance increases a lot using the DFE architecture, the only disadvantage is the store/reading time. Palabras claves : Compression, Dataflow Engine (DFE), FPGA, CPU, DNA, Maxeler