File compression

File compression is important to increase storage density of genomes per hard drive and reduce time to transfer files. We offer two approaches to significantly reduce file size.

QV binning

Compression of per-base quality values (QVs) is an effective method to reduce file size. We adopt per-base QV binning in ccs and can achieve up to 40% reduction in BAM file size with similar SNV and InDel detection performance.

How it works: QV compression is performed on base QVs after read quality rq has been computed. For each consensus read, per-base QVs are assigned to seven bins and fixed average QVs are assigned to each bin

QV bin	Mean QV	ASCII
[ 0, 6 ]	Q3	$
[ 7, 13 ]	Q10	+
[ 14, 19 ]	Q17	2
[ 20, 24 ]	Q22	7
[ 25, 29 ]	Q27	<
[ 30, 39 ]	Q35	D
[ 40, 93 ]	Q40	I

Sorting by sequence similarity

In addition to QV filtering, sorting unaligned HiFi BAM files by sequence similarity can reduce file size by up to an additional 30% using

samtools sort -M -K 20 hifi_reads.bam -o hifi_reads.sorted.bam