For gene level models, do they work with rare variants only (i.e., is it required to set --freqUpper 0.05)?

[ppjeep]

1. In the example folder of rvtest package, you did not specify --freqUpper 0.05 in cmd.sh. Here is the example code. If I use --freqUpper 0.05 in the same command, the gene based results will be different. Is it OK if I run gene based analysis without using --freqUpper 0.05?
   ../executable/rvtest --pheno pheno --inVcf example.vcf.gz --setFile setFile --burden cmc,cmcWald,zeggini,zegginiWald --out out9
2. In the revtests wiki, you mentioned "To extract more than one annotation types, use --annoType 'Start_Gain|Stop_Loss|Start_Loss|Essential_Splice_Site|Stop_Gain|Normal_Splice_Site|Synonymous|Nonsynonymous will extract LOF (loss of function) mutations". Synonymous should not be included in the LOF mutations. Could you give me a reference paper (or resources) to find the definition of LOF (loss of function) mutations? I need some references about LOF mutations for my paper.
   Thanks!

[zhanxw]

1. Usually rare variant test will use a upper frequency limit.
2. I don't have the reference. Actually, I learned the definition of LOF from my colleagues. Maybe @dajiangliu can give you a reference.

[dajiangliu]

I agree that synonymous variants may not be considered lof. For some of our own work, we group variants that are nonsynonymous, stop_gain or splice [see e.g. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5709146/] The choice of the annotation used to group variant is more of a personal taste, no right or wrong.