SNPs and file formats

[dawe]

Hi, I'm looking at some sequence data in order to find mutations. Since I (probably) have to filter the variants with known SNP and indels I would like to have some hint on which file format/tools you are using.
Once I have a list of candidate mutations, I planned to use BEDTools to perform most of the operations. The only problem is that I have different file formats I can operate and this puzzles me:
1- I call variants as result of samtools/samtools.pl. The format is samtools pileup
2- I have snp131 data in BED format (from UCSC genome browser)
3- I guess I can get SNPs from 1000g in VCF format (although I still have to study which is the file(s) I need).

BEDtools should support either BED and VCF format. What about samtools pileup? I guess I may try to convert it in one of both formats...

Puzzle #1 (samtools to BED):
- single nucleotide variations can be converted as CHR:POS-POS+1 (I guess)
- How do I convert small insertions? CHR:POS-POS ?
- How do I convert small deletions?

Puzzle #2 (samtools to VCF):
- Simply, is there a way to directly convert samtools pileup to VCF?

Any suggestion is appreciated :-)
Thanks

d

[epigen]

I had a similar problem with SNPs, but did not consider indels from samtools pileup. For the dbSNP BED file, I converted insertions into POS-1-POS, because BEDTools doesn't find an overlap for features with start=end. I did the same for samtools pileup SNPs to make them zero-based BED as well. The VCF format I don't know.

[dawe]

I had a similar problem with SNPs, but did not consider indels from samtools pileup. For the dbSNP BED file, I converted insertions into POS-1-POS, because BEDTools doesn't find an overlap for features with start=end. I did the same for samtools pileup SNPs to make them zero-based BED as well. The VCF format I don't know.

In the end I've found that samtools 0.1.8 comes with a handy sam2vcf.pl script (which converts to VCF v3.3, I've used vcf-convert from vcf tools to further convert to v4.0). VCF files can be intersected with BED files using BEDTools and, apparently, it works. I mean, at least I've been able to filter my VCF from entries in dbSNP (as BED) :-)
BTW, I had to convert VCF to a 0-based bed-like to use some Useq programs (such as Alleler)... I'm afraid we're far from having some standards here :-(

d