q value in MutSigCV analysis

Question

[wenbinm]

Hi there,

I just used MutSigCV to analyze triple negative breast cancer mutation data from COSMIC. The result it gave me shows there is only one gene whose q value is less than 1.

I notice that in "Comprehensive genomic characterization of head and neck squamous cell carcinomas. Nature, 517(7536), p.576." they also used MutSigCV to analyze COSMIC data and "only 19 genes has q value less than 1".

So what's the point of using q value if nearly all genes have q value equal to 1?

Thank you all!

Answer 1

[h.mon]

So what's the point of using q value if nearly all genes have q value equal to 1?

To avoid getting false positives. You are performing multiple tests (thousands of them, probably), and you expect by chance alone 5% of them will be false positives. q-values corrects for this multiple testing. In general, it is a good practice, but also depends on your aims, as being more strict about false positives also leads to decreased power, i.e., true effects may be deemed statistically non-significant.

Some references:

https://en.wikipedia.org/wiki/False_discovery_rate (has links to several key statistical papers, in case you want to delve on this)

http://www.nonlinear.com/support/progenesis/comet/faq/v2.0/pq-values.aspx

Differential Expression In Rna-Seq Experiment

http://www.biostathandbook.com/multiplecomparisons.html

[wenbinm]

Thank you for your quick response. It really helps!

[GenoMax]

If an answer was helpful, you should upvote it; if the answer resolved your question, you should mark it as accepted.