Execution halted: vcf2bedpe.R

[arnikz]

https://travis-ci.org/github/GooglingTheCancerGenome/CNN/builds/667785627#L762

[lsantuari]

I could:

• open an issue on StructuralVariantAnnotation.
• modify the SURVIVOR simSV VCF file to be read correctly by StructuralVariantAnnotation

It seems to recognize the SURVIVOR simSV VCF file as a Delly VCF file, with the missing CT field.

[arnikz]

That's fine. For time being, a quick fix is to limit the VCF->BEDPE to GRIDSS

https://github.com/GooglingTheCancerGenome/CNN/blob/787d879145e3ad41a164ee07b2285515310daadb/run_local.sh#L16

[lsantuari]

Converting the truth set of the artificial SVs into BEDPE is still necessary to compare the GRIDSS callset to. See here and here, for instance. So this issue is on my priority list, also for #36 and #41.

[lsantuari]

Adding an extra step to treat TRA as CTX is a quick fix at the moment:

sed 's/TRA/CTX/g' htz-sv.vcf > htz-sv_CTX.vcf

and then using htz-sv_CTX.vcf as test.vcf

[arnikz]

Converting the truth set of the artificial SVs into BEDPE is still necessary to compare the GRIDSS callset to. See here and here, for instance. So this issue is on my priority list, also for #36 and #41.

Yes, I want to use test.vcf and gridss.vcf as a quick fix for run_local.sh

[arnikz]

Adding an extra step to treat TRA as CTX is a quick fix at the moment:

sed 's/TRA/CTX/g' htz-sv.vcf > htz-sv_CTX.vcf

and then using htz-sv_CTX.vcf as test.vcf

Let's keep the data as is.

[lsantuari]

As a reference, treating SURVIVOR simSV TRA SVs as TIGRA CTX SVs seems to work.

[lsantuari]

Adding an extra step to treat TRA as CTX is a quick fix at the moment:
sed 's/TRA/CTX/g' htz-sv.vcf > htz-sv_CTX.vcf
and then using htz-sv_CTX.vcf as test.vcf

Let's keep the data as is.

I agree.

[arnikz]

nope
https://github.com/GooglingTheCancerGenome/CNN/blob/ea119778d8b455c26ddf258107e548895597ee05/run_local.sh#L28

[arnikz]

It wasn't a quick fix after all 😉

https://github.com/GooglingTheCancerGenome/CNN/blob/a276f179aae28cfbe6d860e97df32afd1afbe05c/run_local.sh#L28

🙄 Where did DELLY come from? I missed your #43 (comment).

Moreover, diff iss32 (broken) and dev branch (passes) shows change: TRA->BP.

[lsantuari]

It wasn't a quick fix after all 😉

https://github.com/GooglingTheCancerGenome/CNN/blob/a276f179aae28cfbe6d860e97df32afd1afbe05c/run_local.sh#L28

🙄 Where did DELLY come from? I missed your #43 (comment).

If you process the SURVIVOR VCF file with the artificial SVs with breakpointRanges you get the following error:

There were 16 warnings (use warnings() to see them)
Warning message:
package ‘argparser’ was built under R version 3.6.2
Error in .breakpointRanges(x, ...) : Delly variants missing CT:
Calls: breakpointRanges -> breakpointRanges -> .breakpointRanges
In addition: Warning message:
In .breakpointRanges(x, ...) :
  Found 1000 duplicate row names (duplicates renamed).
Execution halted

This is because SURVIVOR, like DELLY, uses the SVTYPE TRA for translocations, instead of BP (GRIDSS) or CTX (TIGRA).

Moreover, diff iss32 (broken) and dev branch (passes) shows change: TRA->BP.

[arnikz]

OK. It's still unclear to me which of the three (TRA, BP or CTX) should be used in this script to pass and/or to resolve merge conflicts in the branches?

[arnikz]

If you process the SURVIVOR VCF file with the artificial SVs with breakpointRanges you get the following error:

Yes, the very same error shown here.

[lsantuari]

OK. It's still unclear to me which of the three (TRA, BP or CTX) should be used in this script to pass and/or to resolve merge conflicts in the branches?

CTX

vcf2bedpe.R works with test.vcf when all the occurrence of TRA are substituted with CTX:

sed 's/TRA/CTX/g' test.vcf

However, you still have to direct the output of sed to an intermediate file that is used in input to vcf2bedpe.R.

[arnikz]

Last commit passes but the code does not make use of the VCF file. Please fork the iss32 instead of dev to continue.

[lsantuari]

Done. Branch iss43 can be deleted.