tool to compare sv between two files

Question

[Medhat]

I called structure variant using Sniffles for two samples (using PacBio reads), I would like to compare the resulted two files to know what is common and different in both files.

what first come to my mind is to use vcf-isec, but I do not know how it will deal with the deletion size if they are exist in the two files but different in length for example, translocation also if its origin is the same but the location transferred to is different! etc ..

vcf-compare could be used but I do not know if it is the right way!

is there is a tool that could do this in an efficient way? anyone have experience with that?

Thanks,

Answer 1

[Zev.Kronenberg]

One simple way to do this is to merge the SV calls and look at which calls collapse. I've writing some code to do this in VCF format.

https://github.com/zeeev/mergeSVcallers

[Medhat]

I merged them before using vcf-merge it gives a clear result when variant exist in one file but absent from the other file, but when there is variant in the same position (same type of variant ex. DUP or different ex. INV, TRV then it become an issue), I will try the tools and give feed back. Thanks.

[Medhat]

First in installation I have this warning:

src/mergeSVcallers.cpp: In function ‘void manageLoopOverVar(std::vector<vcflib::Variant*>&)’:
src/mergeSVcallers.cpp:603:18: warning: comparison between signed and unsigned integer expressions [-Wsign-compare]
             if(i == tmpdata.size()){

and there is some warning and it skipped TRA

WARNING: TRA events are skipped  

WARNING: could not set region: seqid: 1 file: variant1_sort.vcf.gz
INFO: Seqid might not be in file
INFO: sorting: seqid: 1
n SVs in chunk: 7

in this fields:

CIPOS,Number=2, Type=Integer,Description="Confidence interval around POS for imprecise variants CIEND,Number=2, Type=Integer,Description="Confidence interval around END for imprecise variants

I have them from -10 to 10 as seen below

CIPOS=-10,10;CIEND=-10,10

what is this means? (does it means that there is 10 reads supporting this results?)

[Zev.Kronenberg]

It is simply a confidence interview around the start and end of the SV. By default I set it to ten. When multiple SVs are merged it gets wider.