Deltas circulares: proposta e avaliação do impacto da aritmética modular na compressão dos índices de qualidade do sequenciamento de genomas

Abstract

Genomics is an area of genetics that has contributed to several scientific advances, such as the complete sequencing and analysis of genomes of different species. The data generated by the sequencing machines are initially stored in text mode in FASTQ format, which they can reach, for example, about 300GB for a single human cell. Storing and analyzing the sequence of thousands or millions of organisms is a task that requires the efficient use of many computational resources, which validates the importance of compression algorithms for this area. Quality scores (QS) are the most difficult part to compress in genome sequencing files, because their dictionary is extensive and generates a very large number of possible combinations. The objective of this work is to propose a data transformation, the circular deltas (CD), for the quality scores of genome sequencing and to evaluate their impact on the compression of this type of data. This transformation takes advantage of the observation that neighboring quality scores vary little from one to the other and explores the use of modular arithmetic to minimize the representation of these variations. The use of the entropy calculation expresses the number of bits needed to represent each signal, standing out as a fundamental part in the evaluation process. As analyzed from the tests performed, even though the entropy of the QS values were greater than those ND and CD, in the end the compression was favorable to files with QS lines, allowing a compression ratio higher than files with ND and CD lines.