Multiprocess Implementation of DNA Pre-alignment Filtering using the Bit Matrix Algorithm

Abstract

With the emergence of more advance technologies in genomic research, it has becomes easier to answer biological and genomic questions. To be able to analyze advance genomic data, however, the process has to start with the determination of sequence variations between two genomic sequences. This process is called the read mapping process. This involves mapping reads to a similar reference genome which can be computationally intensive. Because of this, most works adapt the filter-verification paradigm. More recently, there are works that focus on improving the filtering phase by adding another layer to it called the pre-alignment filter. This manuscript focuses on the bit matrix pre-alignment filter with a goal of improving the implementation by using multiprocessing techniques that exploit multi-core CPUs. The results of this manuscript show that there is an improvement in using the parallel version over the sequential version by an average factor of 4.2 without sacrificing accuracy.