KLAST: a new high-­performance sequence similarity search tool

KLAST is a fast, accurate and NGS scalable bank-to-bank sequence similarity search tool providing significant accelerations of seeds-based heuristic comparison methods, such as the Blast suite. Relying on unique software architecture, KLAST takes full advantage of recent multi-core personal computers without requiring any additional hardware devices.KLAST is a new optimized implementation of the PLAST algorithm (1), to which several improvements have been made. KLAST is fully designed to compare query and subject comprised of large sets of DNA, RNA and protein sequences using KLASTn, KLASTp, KLASTx, tKLASTx and tKLASTn methods. It is significantly faster than original PLAST, while providing comparable sensitivity to BLAST and SSearch algorithms. KLAST contains a fully integrated data-filtering engine capable of selecting relevant hits with user-defined criteria (E-Value, identity, coverage, alignment length, etc.).KLAST has been benchmarked on metagenomic data sets from the Tara Oceans International Research Project (2). The main goal of the test was to evaluate speedup and quality of results obtained by KLAST in comparison with BLAST, which is usually used at Genoscope to run sequence comparisons. Quality was evaluated in two ways. First, crude results from both tools were compared, i.e. how much results from BLAST are also found by KLAST. Second, by using results from both tools to assign each query to a taxonomy entry. KLAST achieved sequence comparisons up to 18x times faster than BLAST, while covering up to 96% of the results produced by BLAST. This benchmark illustrates the benefits of using KLAST both in terms of quality results and speed on the deciphering of Tara Oceans metagenomic data.To provide users with an advanced sequence similarity search platform, the KLAST engine has been integrated into several software tools, from the command-line up to full-featured graphical data analysis platforms such as ngKLAST, KNIME and CLC bio’s Genomics Workbench. In all cases, the KLAST system provides an integrated algorithm suite that automatically processes analysis workflows that includes similarity searches, hits annotations, and data filtering.