how to identify different snps between two groups?

[slowsmile]

Dear seq members

I am fairly new to SNP analysis. In my study, I have 2 groups, treatment and control, each containing 3 biological replicates. My goal is to find different snps between the two groups and I'm using samtools mpileup | bcftools in my analysis.

Shown below are my codes:

samtools mpileup -C 50 -uDS -q 10 -f MYPATH/human_ref_genome.fa output/tr1.bam output/tr2.bam output/tr3.bam output/ctr1.bam output/ctr2.bam output/ctr3.bam | bcftools view -bvcg - > ./snp_output/snp_cmp.bcf

bcftools view -c -1 -3 ./snp_output/snp_cmp.bcf | vcfutils.pl varFilter -D 100 > ./snp_output/snp_cmp.vcf


Did I use the correct command? Why did the output snp_cmp.vcf file contain millions of snps while I'm expecting hundreds of snps of interest? How did I only limit (filter) my scope to the different snps between the two groups and which program should I use to achieve this goal?

Thanks in advance.

[Xi Wang]

I am also interested in this question. Hope some experts could help.

[gringer]

Why did the output snp_cmp.vcf file contain millions of snps while I'm expecting hundreds of snps of interest?

See here:

http://seqanswers.com/forums/showthread.php?t=16247

Hundreds of thousands of SNPs are expected and perfectly normal. However, I think millions would be pushing it a bit too much for comparing only a few individuals.

In my study, I have 2 groups, treatment and control, each containing 3 biological replicates

Three biological replicates seems quite a low number for SNP frequency studies. I'm more used to hundreds or thousands of individuals compared (e.g. WTCCC). I'd expect you'll have a lot of trouble distinguishing between false and true positive results with only 6 individuals.

Edit: I've just noticed that the thread I linked to was also from you discussing the same issue. Oh well, I guess it must have needed some clarification.

[david.tamborero]

I'm actually newbie in bioinformatics and maybe I am missing something, but as far as I understand you should be interested in those SNPs which are different among the two groups. If so, you could use some of the tools designed for finding somatic mutations.

FYI, now I am getting a try to Varscan. It uses the group1_pileup and group2_pileup files as input and then outputs some nice statistics about the 'somatic status' of any variant.

hope it helps,

cheers
david