metrics on BAM file

[Moip]

Hi everyone,
I am doing exome sequencing, paired end, but not used to analyze reads.
After alignement, how do i proceed the BAM file?
Thank you

[Richard Finney]

How many samples do you have?
How did you align the reads?
Are they tumor and normal samples?
Are looking for germline characteristics or somatic changes?

[Moip]

I have three normal samples and looking for germline characteristics.
Mapping with BWA-MEM.
Thank you for you help Richard

[Richard Finney]

You will probably want to do some variant calling on the bam files. This may include SNPS, fusions, amplification or deletions.

There are many tools to do this. Searching for these tools is straightforward:

https://www.google.com/search?q=germline+variant+calling+on+bam+files

You may want to use favorite, known tools already used at your organization, so ask around. A new tool may require some non-trivial installation.

[Moip]

Thank you for you reply.
i need to understand some details, once i have a BAM file, before variant calling i have to remove duplicate, recalibrate BAM? run other metrics ?
sorry i am completely new to this.

[Richard Finney]

You could do some quality checks and remove duplicates and recalibrate; but if you are just starting off, just go ahead and assume the data is good and try and see if you can get a report on variations using a commonly used tool. After you get that working, you can check out the other stuff and re-run a new workflow. Order of importance would be : QC, duplicate removal or marking , then re calibration. Many here at work say recalibration is not worth the effort. Theoretically marking or removing duplicates may be important upstream to a variant caller, and again, theoretically a variant caller could do that work for you so read the docs for your tools and follow their advice.

Do try and run a QC tool on your fastqs early.
Absolute easiest QC on bam files is "samtools flagstat" (google it).
Study the results and see if you can make sense of them.

[Moip]

Thank you, i ll give a try. Do i ignore UMi for this analysis?