CNV detection using BAM file

Question

[reza]

Hi everyone

i want to detect CNV from BAM file. My question is: in what level i can use BAM file to CNV detection. I mean, unsorted BAM or Sorted BAM or deduplicated BAM (Using Picard) or realigned BAM (using GATK) is better to CNV calling?

[h.mon]

I think first you must choose a method (meaning software), then read its manual and follow its instructions.

[reza]

i want to use FREEC control but my question is not related to used software

[WouterDeCoster]

Of course it is related.

[reza]

no more help to my request?

[ivivek_ngs]

as Eric said, you already have your answer. Now coming to more specific. I suppose you are referring to Control-FREEC and not just FREEC. Try to get hold of the most updated one and not the old one. Valentina Boeva maintains the doc pretty well and all the informations are pretty good there. You your bams need to be sorted. GATK processed BAM files should be working well if you have followed the process properly and not broken the input parameters for Control-FREEC program. You might have to play with the threshold over time to understand the output and minimize the signal to noise ratio and see the real differences. I am speaking from my experience of working with Control-FREEC with WES samples on calling somatic CNVs. I was able to do it after a bit of tweaking and very amazing response from the author of the tool. On other note , my data was not pure and pretty heterogenous tumor, if you have good depth of data and clean in that case even other tools should suffice. Only close tool for my work was ADtex then. Good luck!

[reza]

Thanks vchris_ngs, I will use your advice.

Answer 1

[Eric T.]

Most read-depth-based CNV callers operate on sorted BAM files. For targeted amplicon sequencing, don't remove PCR duplicates. For WGS and hybrid capture, do remove duplicates. The authors of each method will usually describe exactly what you need to provide as input to their programs.

[reza]

Thanks for your clear response Eric.