Objective and Comprehensive Evaluation of Bisulfite Short Read Mapping Tools. (15th April 2014)

• Record Type: Journal Article
• Authors: Tran, Hong; Porter, Jacob; Sun, Ming-an; Xie, Hehuang; Zhang, Liqing
• Other Names: Hong Huixiao Academic Editor.
• Is Part Of: Advances in bioinformatics. Volume 2014(2014)
• Publication Date: 2014-04-15
• Subjects: Bioinformatics -- Periodicals; Bioinformatics; Computational Biology -- Periodicals; Periodicals; 570.285
• DOI: 10.1155/2014/472045
• Languages: English
• ISSNs: 1687-8027

Abstract

Abstract : Background . Large-scale bisulfite treatment and short reads sequencing technology allow comprehensive estimation of methylation states of Cs in the genomes of different tissues, cell types, and developmental stages. Accurate characterization of DNA methylation is essential for understanding genotype phenotype association, gene and environment interaction, diseases, and cancer. Aligning bisulfite short reads to a reference genome has been a challenging task. We compared five bisulfite short read mapping tools, BSMAP, Bismark, BS-Seeker, BiSS, and BRAT-BW, representing two classes of mapping algorithms (hash table and suffix/prefix tries). We examined their mapping efficiency (i.e., the percentage of reads that can be mapped to the genomes), usability, running time, and effects of changing default parameter settings using both real and simulated reads. We also investigated how preprocessing data might affect mapping efficiency. Conclusion . Among the five programs compared, in terms of mapping efficiency, Bismark performs the best on the real data, followed by BiSS, BSMAP, and finally BRAT-BW and BS-Seeker with very similar performance. If CPU time is not a constraint, Bismark is a good choice of program for mapping bisulfite treated short reads. Data quality impacts a great deal mapping efficiency. Although increasing the number of mismatches allowed can increase mapping efficiency, it not only significantly slows down the program, but also runs the risk of having increased false positives. Therefore, users should carefully set the related parameters depending on the quality of their sequencing data.