Crumble: reference free lossy compression of sequence quality values

Motivation The bulk of space taken up by NGS sequencing CRAM files consists of per-base quality values. Most of these are unnecessary for variant calling, offering an opportunity for space saving. Results On the Syndip test set, a 17 fold reduction in the quality storage portion of a CRAM file can be achieved while maintaining variant calling accuracy. The size reduction of an entire CRAM file varied from 2.2 to 7.4 fold, depending on the non-quality content of the original file (see Supplementary Material S6 for details). Availability and implementation Crumble is OpenSource and can be obtained from https://github.com/jkbonfield/crumble. Supplementary information Supplementary data are available at Bioinformatics online.