samtools/mpileup heterozygous SNPs calling

[combiochem]

I'm trying to get heterozygous SNPs from illumina DNA sequencing data.
I've used samtools/mpileup pipeline and have some questions about the options.

There are many posts related in the BAQ calculation. As known, the calculation is default and if the options -B is used, more SNPs could be detected (sacrificing the specificity). I've tested with my samples and the difference is almost ~2x or ~3x difference. (-uf vs -Buf, -Euf didn't make a big difference compared -uf)
Is there anyone with experience with -B -E and -I options?

Actually, we are not interested in INDELs so I thought that the option -I could be used for ignoring INDELs calling.
But when I compared the results (for heterozygous SNPs) with -I and without -I, many detected heterozygous SNPs in the results with -I option are actually the INDELs cases in the results without -I option. So is it better to ignore those SNPs, in another word, should I "not" use the -I option?

I really want to know the appropriate options for calling heterozygous SNPs.

[swbarnes2]

My tiny bit of experience:

A sanger-verifed herterozygous SNP disappeared when I omitted -B, and reappeared when I put it on. I've had other projects, where I didn't sanger verify, but where multiple related projects all had SNPs in the same gene, which was a highly likely candidate, that were virtually uncallable without the -B option.

I wouldn't filter out indels, unless you were using an aligner that you know can't handle them.

[combiochem]

My tiny bit of experience:

A sanger-verifed herterozygous SNP disappeared when I omitted -B, and reappeared when I put it on. I've had other projects, where I didn't sanger verify, but where multiple related projects all had SNPs in the same gene, which was a highly likely candidate, that were virtually uncallable without the -B option.

I wouldn't filter out indels, unless you were using an aligner that you know can't handle them.

Thanks for sharing tips. Actually, the reads were aligned by BWA, so it's better to consider INDELs. Have you used '-E' option (Extended BAQ computation) too?

[swbarnes2]

Thanks for sharing tips. Actually, the reads were aligned by BWA, so it's better to consider INDELs. Have you used '-E' option (Extended BAQ computation) too?

The author of samtools suggested it in a thread where I mentioned my problem with not using -B, but I haven't tried applying to those samples yet.

For my projects, false positives are not that big a problem. I'm looking for candidate phenotype-causing SNPs most of the time, so sanger-checking a modest number of false positives is not a big deal. But I don't want to miss the real deal because the software was overzealous in trying to help me, so it's safer for me to turn it off entirely.

But it's good to know that in your tests, -E worked about as well as -B.

[sijungyun]

For my case, using '-B' increased the number of SNP calls by 2 ~ 5 times from Illumina sequencing data.