statistical analysis of WES

Question

[statfa]

Hi,

I have to work on CNV detection but I am not sure what the next step is after detecting the CNVs. What more analysis do you think can be conducted next?

Thanks in advance

[Eric T.]

Could you tell us what sort of samples you've sequenced? Cancer? Germline samples with rare diseases, common ones, or different sub-types of a family of diseases? Cohorts of related individuals, or unrelated individuals who happen to have some other relevance? Neonatal or adult? What hypotheses are you looking to investigate?

One basic analysis approach would be to treat CNVs as gene-level mutations, where copy number loss is the loss of one or both alleles and copy number gain could increase the gene dosage. Depending on your hypothesis, you might look for recurrent CNVs linked to the disease(s) of interest.

[statfa]

Thanks for your reply. The data are genomic DNAs of adult individuals with rare genetic diseases. Do you have any idea rather than finding CNVs? I'm a beginner, that's why I face these questions. I'm currently reading basic information to comprehend the data.

[Eric T.]

If you're not sure which analyses to run initially, I recommend using a third-party pipeline like bcbio-nextgen. See if you can install and run the standard bcbio pipeline for germline variant calling (this is a learning experience, too), then examine all the outputs and discuss the results with your advisor.

[statfa]

Thank you for your assistance. I will check it out.

Answer 1

[WouterDeCoster]

Based on literature and personal experience the prediction of copy number variants based on exome data is noisy/problematic/troublesome. Probably the first thing you want to do is wetlab validation to be sure that the CNV is real and not an artefact.