GATK RNA-seq best practices

Question

[sbombin]

Hello,

I am using GATK Calling variants in RNAseq: http://gatkforums.broadinstitute.org/gatk/discussion/3891/calling-variants-in-rnaseq

I cannot understand will the expression level will affect the total number of variants found in treatment groups. So for example, if some treatment elevated overall gene expression level, could it affect how many variants I will have in my final vcf file? I asked this question in GATK comments but did not receive answer yet. Thank you for your explanations.

Answer 1

[Devon Ryan]

If you plotted "Statistical power for variant detection" vs. "average coverage of a gene/transcript", you'd find a roughly sigmoidal curve. More reads (from increased expression) means greater ability to find variants. You'll be hard pressed to find variants in very lowly expressed genes (note that you need to consider this if you're trying to make genotype:phenotype comparisons).