Fig. 3

Plot of runtime vs number of query sequences in a BLASTN XML (−outfmt 5) results file. The graph is a linear-log plot, the number of query sequences is shown on a log scale, due to the exponential nature of runtime data and the large numbers of sequences involved. Each of the scripts were run against BLAST XML files containing 10³, 10⁴, 10⁵, 10⁶ query sequences respectively. We did not include a run of 10⁷ XML query sequences as the file size became impractical for our system, taking up all 128Gb of ram (this resulted in a outOfMemoryException on the Java parser). All versions of BLAST QC were run using default parameters (no command line options specified), which is to order hit sequences for each query by e-value. Fichot’s BioPerl script was also run using default parameters with no thresholds implemented. Both the BioPerl script by Xiaodong Bai and the Python script by Peter Cock only function as XML (outfmt 5) to tabular format (outfmt 6) converters, so no input parameters are required