SNPs detection by eBWT positional clustering

Table 1 Pre-processing comparative results of ebwt2snp (i.e. building the eBWT using either eGSA or BCR) and DiscoSnp++ (i.e. building the de Bruijn graph)

Dataset	Coverage per sample	#reads	Preprocessing	Wall clock^a (h:mm:ss)	RAM (MB)
HG00096 (ch. 22)	29×	15,000,000	gsacak	0:53:34	100,607
			eGSA	1:41:37	30,720
			BCR	4:18:00	1,970
			DiscoSnp++	0:01:09	5,170
HG00100 (ch. 16)	22×	20,000,000	gsacak	1:13:04	112,641
			eGSA	3:39:04	30,720
			BCR	6:10:28	3,262
			DiscoSnp++	0:02:01	6,111
HG00419+NA19017 (ch. 1)	43×–47×	93,657,983	BCR	105:28:30	73,977
HG00419+NA19017 (ch. 1)	43×–47×	93,657,983	DiscoSnp++	0:32:37	621

Wall clock is the elapsed time from start to completion of the instance, while RAM is the peak Resident Set Size (RSS). Both values were taken with /usr/bin/time command. Note that for the last collection we have used a variant of BCR that keeps the \({ \textsf {ebwt}} (\mathcal {S})\) in internal memory. eGSA and gsacak have not been tested on the last dataset since they required too much disk space and RAM, respectively
^aWe recall that DiscoSnp++ makes use of multiple cores while ebwt2snp is currently designed to use one core only, thus explaining the difference in speed

ISSN: 1748-7188