|Thread||Thread Starter||Forum||Replies||Last Post|
|Mpileup/BCFtools pipeline not picking up indels (suggestions please)||cam.jack||Bioinformatics||7||05-17-2013 01:05 PM|
|Pipeline to simulate short reads and call indels||pmaugeri||Bioinformatics||17||03-01-2012 03:45 PM|
|samtools indels||Robby||Bioinformatics||3||11-08-2011 07:02 AM|
|samtools tview produces "Floating point exception" on big file?||pmaugeri||Bioinformatics||2||10-28-2011 08:06 AM|
|indels from samtools||coonya||Bioinformatics||2||06-06-2011 04:12 PM|
|06-15-2011, 09:51 AM||#1|
Location: Baltimore MD
Join Date: Jun 2011
Samtools pipeline produces many indels
I'm new to the community and just beginning to understand many of the programs used for SNP selection. In my current exome seq project, I am looking for homozygous SNPs in a consanguineous pedigree with multiple affected siblings. My pipeline is as follows:
previously aligned input.bam files were obtained from sequencing facility
$ samtools sort input.bam input.sorted.bam
$ samtools rmdup –s input.sorted.bam input.sorted.rmdup.bam
$ samtools index input.sorted.rmdup.bam
$ samtools faidx HG19.fa
$ samtools mpileup –uf HG19.fa input.sorted.rmdup.bam > variants.raw
$ bcftools view –bvcg variants.raw > variants.raw.bcf
$ bcftools view variants.raw.bcf | vcfutils.pl varFilter –d 3 –D 1000 –G 20 > variants.flt.vcf
Afer this, I discard common variants using the 1000 genomes and/or dbSNP
I then grep for homozygous variants, variants shared among affected individuals and found in the heterozygous state in the unaffected parents.
This protocol seems to be generating a large number of INDELs (>50%) compared to SNPs. Is this unusual? Should I have more stringent filters in place?
Also, once I have obtained a final list of variants, what programs are recommended currently for functional analysis?
|06-15-2011, 10:13 AM||#2|
Location: San Diego
Join Date: May 2008
samtools mpileup –Buf HG19.fa input.sorted.rmdup.bam > variants.raw
In my experience, the BAQ calculations eat SNPs. -B will turn those calculations off.
If you compare the two pileups with and without the BAQ, what you may see is that the BAQ calculations are drastically dropping the quality scores of real SNPs, causing the SNP caller to ignore them, due to low quality. I believe this is an attempt to reduce false positives due to indels.
|06-17-2011, 04:30 AM||#3|
Join Date: Feb 2008
Finding indels is much harder, no one so far really understands that. Samtools is calling many false indels, and the existing database is lacking many true indels.
As to BAQ, if you really care about sensitivity, disable it. For general purposes, use it. BAQ trades a couple percent sensitivity for hugely improved specificity.
|analysis, filtering, functional, indels, samtools|