I was doing the variant calling of whole exome sequences attained from tumor tissues. I used an inbuilt loop for filtering it upto to the formation of vcf files. The loop ran successfully and gave the resultened files. But when I created the TMB plots out of it the graph looks linear with no similar TMB for the all samples that have been used. could you please help to identify what might have went wrong

Hi bioinffun,

The linear TMB plot with no clustering suggests your variant calls may lack sample-specific discrimination, likely due to filtering applied uniformly across all samples in the loop without matched normals or proper tumour-only safeguards. Common issues include over-filtering germline variants (especially in exomes with poor population databases), failing to exclude FFPE artefacts, or not using panel-of-normals (PoN) for tumour-only calling. Check if your loop used --normal_panel in Mutect2 or equivalent; without it, shared artefacts inflate TMB identically across samples.

Quick diagnostic in R:

vcf_files <- list.files(pattern = "*.vcf.gz")
tmb <- sapply(vcf_files, function(v) {
  vcf <- readVcf(v, "hg38")
  sum(info(vcf)$TUMOR_AF > 0.05 & info(vcf)$FILTER == "PASS")
})
plot(tmb, pch=19, ylab="TMB (mut/Mb)", xlab="Sample index")

If values are near-identical, re-run with tumour-normal pairs or add --germline-resource and --panel-of-normals.

Kevin