SNP base calling

[shuang]

My SNP data is from Sanger sequencing. Multiple samples cover varied regions, not necessary the same fragments. I performed alignment via bwasw and pileup via samtools.

What would be the differences between
1. pileup multiple samples all together
2. pileup one sample at a time?


Also, would QUAL score, DP, AC be affected dramatically?

[volks]

mpileup doesnt give you an output at positions where there is no coverage. so if you want to compare different samples it might be more convenient to generate the pileup together.

[shuang]

I actually do not need to know the position when a sample doesn't cover that base.

Other than that, any else would make differences?

[swbarnes2]

If a SNP is called in one sample, and not another, it is helpful to look at the other sample, to determine if that other sample really is wt, or if coverage was just too low for it to make the same SNP call, and doing mpileup together helps for that. Unfortunately, what's really helpful is the DP4 values, and mpileup combines them all, which can make it harder to assess the likelihood of each sample. Yes, you get a GQ, but the coverage is helpful as well.

[shuang]

I also notice that QUAL score tends to be much lower in one sample analysis than multiple samples analysis. Why is that way?

[lh3]

If you want to compare between samples, always pool samples together (i.e. generate mpileup across all samples). Mpileup skips sites where there is no coverage across ALL samples. On the other hand, swbarnes2 has the point that DP4 is combined. If you need that information, you may pool samples at first to find sites you are interested in and then run single-sample pileup to get DP4.

[shuang]

When I tried to pileup multiple samples together, even a sequence/read did not cover the SNP base was shown as het (1/0). That confused our conclusion.

How do I avoid that? Or how do I tell a het means a real one or means no-coverage?

Also, how do I use DP4 value?

[aslihan]

How to use dp4 values ??

How to use dp4 values ??