Recieving low distribution values from exome data

[ghosholivia]

Hi @brentp ,

For 14 exome samples with approx 3.3GB exome BAM each, I am receiving low distribution values of depths from mosdepth tool compared to other tools such as Qualimap. [On average: 26% low value from the other tool] I have used the parameter mentioned in your example: mosdepth --by capture.bed sample-output sample.exome.bam For both the tool, capture.bed = 12MB

Can you guide me on why there's much difference in output?

Thanks, Olivia

[brentp]

Hi Olivia, what file are you looking at that's showing low distribution values?

[brentp]

I can likely help if you can start by answering this question ^

[ghosholivia]

Hi Olivia, what file are you looking at that's showing low distribution values?

Thanks for your reply. @brentp Those are hiseq-2000 human exome data of ~120 million reads.

Thanks, Olivia

[brentp]

I mean what mosdepth file

[ghosholivia]

I mean what mosdepth file

The file “output.mosdepth.regions.dist.txt”. I ran the python script (plot.dist.py) on this file to get the mean coverage.

[brentp]

ok. that's the right file. now can you clarify what you saw that was unexpected? by that, i mean, can you expand on this

I am receiving low distribution values of depths from mosdepth tool compared to other tools such as Qualimap. [On average: 26% low value from the other tool]

what value are you looking at from mosdepth and how did you get 26% and what is "low"?

[ghosholivia]

Yes. What I meant to say is the mean coverage I’m getting from mosdepth is ~26% lower than the value from Qualimap.

Whereas in WGS depth output, values are not varying as such.

I’m attaching one screenshot of the excel sheet with the differences mentioned above.

[brentp]

ah. ok. the coverage from mosdepth is 26% less than from qualimap. I don't know how qualimap works, but you could try running mosdepth with --fast-mode and see if the numbers match more closely.

If they do, that means that qualimap does not adjust for overlapping read-pairs (mosdepth does by default).

[ghosholivia]

ah. ok. the coverage from mosdepth is 26% less than from qualimap. I don't know how qualimap works, but you could try running mosdepth with --fast-mode and see if the numbers match more closely.

If they do, that means that qualimap does not adjust for overlapping read-pairs (mosdepth does by default).

Okay sure. I’ll try that parameter and check. Thanks a lot.

[aspitaleri]

ah. ok. the coverage from mosdepth is 26% less than from qualimap. I don't know how qualimap works, but you could try running mosdepth with --fast-mode and see if the numbers match more closely.

If they do, that means that qualimap does not adjust for overlapping read-pairs (mosdepth does by default).

Hi first of all great tool! I discovered it and it is very Swiss knife! I found me too the same behavior in comparison with samtools and pileup.sh (from BBMap) and --fast-mod "fix" the discrepancy indeed. Wondering which value is correct to be considered, with or without overlapping corrections? Thanks

[brentp]

Depends what you mean by "correct". If you use --fast-mode then you are double-counting overlapping reads from the same fragment. So you got one piece of information (the fragment) and you counted it twice wherever the reads overlapped.