SRAssembler: Selective Recursive local Assembly of homologous genomic regions

Abstract

Background: The falling cost of next-generation sequencing technology has allowed deep sequencing across related species and of individuals within species. Whole genome assemblies from these data remain high time- and resource-consuming computational tasks, particularly if best solutions are sought using different assembly strategies and parameter sets. However, in many cases, the underlying research questions are not genome-wide but rather target specific genes or sets of genes. We describe a novel assembly tool, SRAssembler, that efficiently assembles only contigs containing potential homologs of a gene or protein query, thus enabling gene-specific genome studies over large numbers of short read samples. Results: We demonstrate the functionality of SRAssembler with examples largely drawn from plant genomics. The workflow implements a recursive strategy by which relevant reads are successively pulled from the input sets based on overlapping significant matches, resulting in virtual chromosome walking. The typical workflow behavior is illustrated with assembly of simulated reads. Applications to real data show that SRAssembler produces homologous contigs of equivalent quality to whole genome assemblies. Settings can be chosen to not only assemble presumed orthologs but also paralogous gene loci in distinct contigs. A key application is assembly of the same locus in many individuals from population genome data, which provides assessment of structural variation beyond what can be inferred from read mapping to a reference genome alone. SRAssembler can be used on modest computing resources or used in parallel on high performance computing clusters (most easily by invoking a dedicated Singularity image). Conclusions: SRAssembler offers an efficient tool to complement whole genome assembly software. It can be used to solve gene-specific research questions based on large genomic read samples from multiple sources and would be an expedient choice when whole genome assembly from the reads is either not feasible, too costly, or unnecessary. The program can also aid decision making on the depth of sequencing in an ongoing novel genome sequencing project or with respect to ultimate whole genome assembly strategies.