This tool extracts a subset of reads where the size of the subset is a percentage of the input size. Sampling is done in a pseudo-random way, which does not guarantee that the extracted subset comprises an exact percentage of the input reads. The input reads can be provided in both interleaved and non-interleaved format and reads marked as paired are kept together.
Read sampling can be useful for reducing coverage of datasets with a very high coverage (>500x coverage) in preparation for a de novo assembly. A reduction in coverage makes the assembly run faster and reduces the chance of having overlapping errors in the reads, thus increasing the assembly quality.
See Options for full usage information.