Meerkat for identification of somatic structural variations for paired samples

Question

[haiying.kong]

---

Apparently, I did not read through the manual carefully.

somatic_sv.pl has some parameters that provide information for normal genome.

---

I am trying to identify somatic structural variations with WES from paired normal and tumor samples by using the software Meerkat. http://compbio.med.harvard.edu/Meerkat/

(1) I read through their manual, but cannot figure out how "somatic" is decided. My understanding is that the software pipleline runs on paired normal sample and tumor sample respectively with slightly different parameter settings and get 2 lists of structural variations. Somatic structural variations are decided by comparing these 2 lists, and this part is not included in the software. I am worrying that the list from tumor would not include all or at least the most of variations in the list from normal. Does any one have any experience with this software? How should I proceed to get somatic structural variations for paired samples?

(2) Could any one please give me the link to download repeat mask file? -R STR /path/to/repeat_mask_file, required, can be downloaded from UCSC

Answer 1

[haiying.kong]

I think the rmsk data is: rmsk.txt.gz on the link: http://hgdownload.cse.ucsc.edu/goldenpath/hg19/database/

Answer 2

[gdavidson]

Hello,

I've used meerkat in the past. Somatic variations are indeed determined by comparing your tumor and normal samples. The script to do this called "somatic_sv.pl" is included in the software in the "scripts" directory. I seem to remember the usage is detailed in the pdf manual where they give examples to perform multiple filtering steps on variants called from your tumor sample.

From my notes they give 4 steps to filter somatic variants:

• Discard events with breaking point within a certain distance of a breaking point from a variant called in the control sample.
• Discard events that are supported by discordant read pairs that are also in the control sample.
• Discard events that are supported by chimeric reads that are also in the control sample.
• Discard events that are supported by non-uniquely mapped reads that are also in the control sample.

You have parameters you can tweak for all filtering steps. Hope this helps.

[haiying.kong]

I am trying to filter variants by following the recommended steps on the manual. Our data is WES for paired samples of normal and tumor. Average depth is 60.

If I run the following commands they all give empty output files without any errors. I checked all input files, and they are correct. Tweaked the parameters a bit, but output is always empty.

perl ${Meerkat}/somatic_sv.pl \ -i ${Meerkat_dir}/Tumor/${Tumor}/${Tumor}.variants -o ${Filtered_dir}/${Tumor}/${Tumor}_somatica.variants \ -F ${Normal_Discord_dir} \ -R ${RMSK} \ -l 1000

perl ${Meerkat}/somatic_sv.pl \ -i ${Meerkat_dir}/Tumor/${Tumor}/${Tumor}.variants -o ${Filtered_dir}/${Tumor}/${Tumor}_somatica.variants \ -F ${Normal_Discord_dir} \ -R ${RMSK} \ -x 1

[Albert Shih]

Hi, I'm also trying to detect somatic SV from our WES data (paired sample too), and I also get empty output files without any errors when execute somatic_sv.pl with its 60-80X parameters. Do you know how to solve it yet? Thask for your help.