Exploring parameter space with hard filters (VariantFiltration)
I am doing an analysis on chimp variants, and do not have enough variants to train the model in order to perform VQSR. I am using hard filters at this point, as per the recommendation. I keep reading that the recommended cutoff values are only recommendations and that I should expect to tweak the parameters further. However, how do I do this in an effective manner? I'm just not sure how to approach exploring the parameter space.
Should I try out different values around the recommended cutoffs and, down the line, analyze how many mutations I get/calculate certain values for quality control and check that they agree with what is expected (CpG transition rate, and so on)? Is there a way to inspect the data to see whether it is behaving well, or do I have to wait until down the line to see how things are going? Any insight into this would help tremendously.