Evaluation of variant calling tools for large plant genome re-sequencing

BMC Bioinformatics

Table 2 Missed and unique calls for seven variant calling tools to show discordance

		FreeBayes	GATK	Samtools	Platypus	SNVer	VarScan	VarDict
Missed calls
Bowtie2	raw	105	73	397	295	521	4000	52,215
	trim	98	79	382	281	526	4146	52,885
	rep1	43	54	204	148	316	5179	45,663
BWA-	raw	254	354	100	1051	1035	14,773	120,053
mem	trim	249	370	84	1044	1073	15,334	120,792
mem	rep1	159	214	47	677	860	20,054	97,722
BWA_raw loci	chr2B93040165	missed	868, 32	225, 47	1019, 37	67	47	230, 66
	chr7D397560208	67, 7	29,4 (f)	94, 7	71,4	7	7	112, 7
	chr2A36897168	220,27	323, 18	44, 36 (f)	457, 21	26	36	200, 45
	chr1A14084557	260, 46	903, 25	134, 45	missed	51	45	207, 42
	chr3B133439650	122, 8	71, 3	114, 8	147, 3	missed	8	120, 8
	chr1A11663841	150,5	79,4	110,4	117,5	5	missed	79,4
	chr1A42695010	336,16	71, 4	152,9	54,9	16	9	missed
Unique calls
Bowtie2	raw	398,709	3568	33,865	1516	1738	124,696	170,880
	trim	389,184	3601	33,522	1790	1688	124,154	170,837
	rep1	185,984	5580	9162	1477	608	78,446	109,310
BWA-mem	raw	472,999	20,057	134,717	6614	4429	30,120	102,023
	trim	447,527	20,397	132,482	7411	4283	29,518	100,868
	rep1	167,601	32,299	36,730	6532	1569	15,526	55,835
BWA_raw loci	chr1A23072417	111,4	21,2	59,2	none	none	none	none
	chr2A178387501	none	101,3	63,2	none	none	none	none
	chr1A286123173	none	none	95,4	none	none	none	none
	chr1A357077575	none	none	none	35,3	none	none	none
	chr7D310148693	6,23	none	10,11	none	19	none	none
	chr1A232066481	none	none	none	none	none	6	none
	chr1B23898050	none	none	none	none	none	none	73, 4

The missed calls were only called by one of the seven variant calling tools. The total missed call numbers of seven variant calling tools using either Bowtie2 or BWA-mem mapping and seven locus examples (five “missed” SNP calls and two falsely missed calls due to filtering) were presented. The falsely missed loci were indicated by call QUAL, total read depth, filtered (score, depth, and f). Other calls were presented by call QUAL score and total read depth (score, and depth). SNVer and vascan did not have QUAL score. The unique calls appeared in only one out of the seven variant calling tools. Total unique calls and seven locus examples were presented. The loci were not called with “none”, or called with QUAL score and total read depth (score, depth). Three unique call loci from one of the FreeBayes, GATK, or SNVer were filtered in other variant calling tools

ISSN: 1471-2105