As far as I am concerned, the standard coverage 50x for the analysis of highly heterogeneous tumor samples such as melanoma is insufficient and scientific literature recommends base coverage around 500x or even 1000x. However, with that coverage the cost of analysis increase enormously.
There is no one coverage that is right or wrong, better or worse. The answer will almost solely depend on the question you are trying to ask of the data and what confidence level you require for detecting SNVs. You would likely require lower coverage if you were only interested in detecting more commonly mutated genes in melanoma, such as BRAF, vs. the much higher coverage required for discovering novel SNVs if for example one were interested in doing a study on tumour evolution or spatial/temporal heterogeneity.
I would agree with Pawel on this. It depends mostly on what question you are asking of the data. But having said that, the purpose for which you perform this analysis will also decide your choice.
For clinical grade analysis, higher the coverage, better your confidence of variant calls (well, almost!). But if it is for exploratory analysis, research interest or experimental reasons, 100x coverage should suffice - purely considering the cost involved and the number of samples you plan to interrogate. Melanoma being a hot pancake in terms of exome analysis in the past few years, you should be fine with this unless you are sniping for rare & low frequency variants.
OK, thank you for all the answers and suggestions. We have finally decided to run our analysis with 100x.
- Badges
- Science method