How to Assess Paired Sample Data from WES

Question

[haiying.kong]

I have paired samples from normal and cancer tissues, and went through whole exome sequencing. I would like to evaluate how much likely the samples are correctly paired, in other words, there is no sample swapping. Could any one please recommend methods or software tools?

[GenoMax]

Not the first thing one should be worrying about when analyzing such data :-)

[ivivek_ngs]

I actually do not favor the part of sample swapping then that would be really a slack at the part of the person making the libraries and then tagging stuffs to be run in sequencer but then the tumor samples might be heterogenous which might have high stromal contamination and that might as a matter of fact interfere with the variant calling downstream , in that case it is worth to do some sort of testing before the variant calling is done to put a fraction of such purity estimation in tools like mutect2 and varscan2

Answer 1

[Lemire]

This is an answer to the question you actually asked:

Look up http://genome.sph.umich.edu/wiki/VerifyBamID

[ivivek_ngs]

This is indeed amazing to see that such a tool existed, obviously the genotype information is needed else the OP cannot carry forward but then computationally such resource availability is only a sign of how manual laborious work can be automated and people can really check for the errors done at manual expense.

[haiying.kong]

Thank you very much. I am looking at the software.

The usage of the software is: verifyBamID --vcf [input.vcf] --bam [input.bam] --out [output.prefix] --verbose --ignoreRG

I am reading the manual, but it is still not clear to me. Is the .vcf file for the SNPs in one sample, and bam file is from the other sample which is paired with the first sample?

Answer 2

[ivivek_ngs]

You can use AbsCN-seq , PurityEst and THetA2 tools to do the following.

Another way to do check the mutant allele frequency of the driver gene. Take a look at this thread

There is also a list of tools available in this link here which can be used both in WGS or WES data.

[haiying.kong]

I think I am asking silly question, but how do we connect purity and sample swapping?

[GenoMax]

Can you specify if you are worried about swaps of normal/tumor for same individual or across samples? Hopefully first.
Ideally you would have independent data (e.g. genotype array) that you can use as a reference for comparison.

[ivivek_ngs]

To be frank I was not concerned about the sample swapping , that could be revealed I believe during the preparation of the library and then charging samples in the sequencer, so during that process each samples are tagged with some barcodes and that back tracking could actually lead to the understanding. However I was more concerned if your tumor is impure or not which while downstream might lead to false positive variants or even false-negatives. So I was mentioning the above. There is a nice article from broad which does at the level of automation some check in order to reduce the effect or nullify the sample swapping. Take a look at here.

The idea of purity is entirely a different thing. It means while your cells were being sorted and then prepared the picking might have lot of normal cells from adjacent tissues which might not entire give us a homogenous tumor culture and thus the stuff sent for sequencing might end up having a lot of stromal cell contamination along with tumor cells , which will be impacting the variant calls later.