Comparison of Burrows-Wheeler Transform-Based Mapping Algorithms Used in High-Throughput Whole-Genome Sequencing: Application to Illumina Data for Livestock Genomes ¹

Rapidly evolving sequencing technologies produce data on an unparalleled scale. A central challenge to the analysis of this data is sequence alignment, whereby sequence reads must be compared to a reference. A wide variety of alignment algorithms and software have been subsequently developed over the past two years. In this article, we will systematically review the current development of these algorithms and introduce their practical applications on different types of experimental data. We come to the conclusion that short-read alignment is no longer the bottleneck of data analyses. We also consider future development of alignment algorithms with respect to emerging long sequence reads and the prospect of cloud computing.

A ubiquitous and fundamental step in high-throughput sequencing analysis is the alignment (mapping) of the generated reads to a reference sequence. To accomplish this task, numerous software tools have been proposed. Determining the mappers that are most suitable for a specific application is not trivial. This survey focuses on classifying mappers through a wide number of characteristics. The goal is to allow practitioners to compare the mappers more easily and find those that are most suitable for their specific problem.

The genome sequence was obtained from 270 sires used in the Germplasm Evaluation (GPE) project. These bulls included 154 purebred AI sires from GPE Cycle VII breeds (Hereford, Angus, Simmental, Limousin, Charolais, Gelbvieh, and Red Angus), 83 F crosses of those breeds, and 33 AI sires from 8 other breeds. The exome capture sequence targeting coding regions of the genome was obtained from 176 of these bulls. Sequence reads were mapped to the UMD 3.1 bovine genome assembly; a mean of 2.5-fold (x) coverage per bull was obtained from the genomic sequence, and the targeted exons were covered at a mean of 20.0x. Over 28.8 million biallelic sequence variants were detected where each allele was present in at least 3 different bulls. These included 22.0 million previously reported variants and 94.1% of the 774,660 autosomal and BTA X SNP on the BovineHD BeadChip assay (HD). More than 92% of the variants detected in targeted exons were also detected from the low-coverage genome sequence. Less than 1% of the variants detected from the combined genome and exome sequence occurred in annotated protein-coding sequences and 5' and 3' untranslated regions (UTR) surrounding the 19,994 annotated protein coding regions. Variation was detected in the coding sequence or UTR of 96.8% of the genes: loss-of-function variants were predicted for 3,298 genes, 14,973 contained nonsynonymous variants, 11,276 had variation in UTR, and 17,721 genes contained synonymous variants. Minor allele frequencies (MAF) were <0.05 for 47.8% of the coding sequence and UTR variants, and MAF distributions were skewed toward low MAF. In contrast, 11.1% of the HD SNP detected in these bulls had MAF < 0.05, and the distribution was skewed toward higher MAF. Genes involved in immune system processes and immune response were overrepresented among those genes containing high MAF loss-of-function and nonsynonymous polymorphisms. Detected variants were submitted to the National Center for Biotechnology Information genetic variation database (dbSNP) under the handle MARC, batch GPE_Bull_GenEx.