Dear Biostars community,
I'm currently working with cohort cancer data to identify genes that are potential drivers. One of the tools I'm using is GRIN as it allows users to combine multiple sources of data. The source of my data are WES paired tumor-normal samples, from which I've computed LOH, CNV, and somatic mutations. When I include all three forms of data with the model I notice that the p and q values are extremely low, about ~10% of genes are considered significant.
Do these q-values make sense, or does the software expect the data to be from WGS? My concern is that the origin of the data changes the null model, though I don't see the option to inform GRIN that the data is from WES.
Something else I've noticed, is that the low q-values appear to be primarily driven by CNV and LOH events, which explains why I see olfactory genes as some of the most significant hits.
I'm open to any suggestions and / or trying new software.