what is DP4 in VCF (Variant Call Format) and AF1?

[gcrdb]

Hi,
Anyone knows what is DP4 in VCF? (I got from mpileup->bcftools view->)
In this VCF line:
chr1 1467416 . G C,T,X 35 . DP=105;AF1=0.5;CI95=0.5,0.5;DP4=36,33,30,5;MQ=47;PV4=0.001,3.5e-05,3.2e-21,0.0073
I undersatnd that:
DP=105 means Depth (or Coverage) is 105.
then what is DP4? DP4=36,33,30,5
I counted number of A/C/G/C , total is 105 but only about half A half C:
(see this pileup)
chr1 1467416 G G 62 0 49 105 .$C$C$C.CC,CCCCC.C.C.CCCC.C.C.C.CCCCCC,T..C...,.CCC.,
.,..,Ct,...,.c,c.cc,,,,,,,,.,,,,.,,,,.,..,.,.,,.,.,,^].^].^]. D!0,2))I3+8)3F+H%44'#2I4II-III22I4II7,I0IIIIFIIIIII/II03%HI<0III2IIIIHI%I#FIII)I1III56IIIIIIIBII/I6IE7@EE




thanks a lot!

[iansealy]

There's a definition of DP4 on http://samtools.sourceforge.net/mpileup.shtml

[gcrdb]

thanks, I saw the definition of DF4, but still not solving my problem:
for example: how many C and G in this line?

chr1 10725 . A C,G,X 34 . DP=17;AF1=1;CI95=1,1;DP4=0,0,17,0;MQ=18 PL 67,45,72,67,0,29,45,66,7
2,67

[Tally]

DP4 refers to the reads covering the reference forward, reference reverse, alternate forward, alternate reverse bases. Eg DP4=2,3,0,0 means 2 reads are the reference base, forward strand, 3 reference base, reverse strand, and no covering reads have an alternate at that position. The sum of DP4 will not always equal DP due to some reads being of too low quality.

[wetSEQer]

thanks, I saw the definition of DF4, but still not solving my problem:
for example: how many C and G in this line?

chr1 10725 . A C,G,X 34 . DP=17;AF1=1;CI95=1,1;DP4=0,0,17,0;MQ=18 PL 67,45,72,67,0,29,45,66,7
2,67

I think for that reason, you can generate the pileup without -g or -v flag and look at that loci to see how many C and Gs.

[lethalfang]

DP4 refers to the reads covering the reference forward, reference reverse, alternate forward, alternate reverse bases. Eg DP4=2,3,0,0 means 2 reads are the reference base, forward strand, 3 reference base, reverse strand, and no covering reads have an alternate at that position. The sum of DP4 will not always equal DP due to some reads being of too low quality.

Anyone knows exactly what constitutes "high quality" in DP4?

[dpryan]

There's no strict definition of high quality, it'll depend on the program used to call the sites (there's no guarantee that a DP4 field will even exist in any given VCF file).

[lethalfang]

There's no strict definition of high quality, it'll depend on the program used to call the sites (there's no guarantee that a DP4 field will even exist in any given VCF file).

What about samtools | mpileup | bcftools in particular?
I know mpileup has a default base quality of 13 and mapping quality of 0, but that's "raw depth" right?
What does samtools' high-quality mean?

[dpryan]

You'd have to go through the source code to find out, I've never seen the specifics described anywhere.