removal of SNP hotspots or clustering of SNPs

[clarissaboschi]

Dear all.

I have millions of SNPs detected by sequencing.
I did a filtration with different criteria and I have a good set of SNPs. I need to submit all the filtered SNPs to NCBI database (dbSNP), but they asked me to check some problematic regions, for example regions with high density of SNPs = >10 SNPs in 50 bases.

I have ~2,000 of those regions with high density SNPs. I have checked manually some of them, and some look ok, but some have low quality indels, repeated regions, homopolymers, etc.

Is it a good idea to remove these high density regions of SNPs? It is impossible to check region by region manually...

Some authors say that "Clustering of SNPs can be a result of the mis-alignment of reads because of the presence of the indels (insertions or deletions) at the beginning or end of reads" (http://www.biomedcentral.com/1471-2164/15/307)

But maybe these regions can be important to be removed....

thanks a lot
Clarissa

[sarvidsson]

I have ~2,000 of those regions with high density SNPs. I have checked manually some of them, and some look ok, but some have low quality indels, repeated regions, homopolymers, etc.

What variant caller did you use? Low quality indels, repetitive regions and homopolymers can be dealt with variant filtering, typically.

[clarissaboschi]

I used samtools/mpileup and different filter parameters.
I am wondering if it is a good idea to use SNP cluster as a filter as well

Example remove snps when- 3 SNPs/10 bases as suggested in the paper http://www.biomedcentral.com/1471-2164/11/469

thanks
Clarissa