Somaticsniper And Filtering With Included Perl Scripts

Question

[DoubleD]

Hello,

I was looking into filtering our Sniper calls using the included perl scrips in order to look at certain mapping quality and somatic score cut-offs. In doing so I ran into a problem. SomaticSniper needs Samtools 0.1.6 to run and says that it will not work with newer versions. To process files for the perl scripts, I tried to get bam-readcount installed with that version of samtools, but ran into errors (not with cmake, but with definitions). I don't remember the error exactly, but it wasn't with sam.h or cmake, etc. I ended up installing samtools 0.16 and bam-readcount installed fine.

My question is, how do these two programs co-exist, or do they? I'm grateful for the provided perl scripts, but if they require a newer samtools than sniper can use, then do you have to just run them on a separate machine?

Thank you, DD

Answer 1

[ernfrid]

You are correct that they (unfortunately) need to compile against different versions of samtools. You shouldn't need separate machines to use these though. If you download both versions of samtools' source code and compile them (you don't need to install them, just compile), then you should only need to set your $SAMTOOLS_ROOT variable to point to the correct version for the program (sniper or bam-readcount) you are installing. For example,

1. Download and compile samtools-0.1.6
2. export SAMTOOLS_ROOT=/path/to/where/samtools-0.1.6/is/compiled
3. Follow directions in compiling SomaticSniper
4. Download and compile samtools-0.1.16 (or some higher version)
5. export SAMTOOLS_ROOT=/path/to/where/samtools-0.1.16/is/compiled
6. Follow directions to compile bam-readcount

[DoubleD]

Wonderful, thank you. I'll make a script to export SAMTOOLS_ROOT to the correct version and then start either program, respectively.

[ernfrid]

To be perfectly clear, once you've compiled the programs you shouldn't need to worry about samtools versions at all. The samtools version is an issue at compilation only.

[DoubleD]

Very clear, thank you