Snakemake corrections

[0xaf1f]

minor bugs in the workflow that would cause it to error out in some cases.

[0xaf1f]

Thanks for merging. There's one more outstanding issue-- this code passes the same GATK jar file to both SomaticSeq and Mutect2, but SomaticSeq cannot work with GATK4. This problem would go away if you make that happen. Otherwise, it's probably necessary to have a gatk3 config setting as well as a gatk4 setting rather than just one for both. Do you think it's worth making this split now, or is an update to GATK4 in the SomaticSeq wrapper coming soon?

[lethalfang]

I'll see if I can switch out GATK3 to GATK4 (i.e., GATK3's CombineVariants to GATK4's MergeVcfs). Best,

On Thu, Jan 25, 2018 at 3:36 PM, Afif Elghraoui notifications@github.com wrote:

Thanks for merging. There's one more outstanding issue-- this code passes the same GATK jar file to both SomaticSeq and Mutect2, but SomaticSeq cannot work with GATK4. This problem would go away if you make that happen. Otherwise, it's probably necessary to have a gatk3 config setting as well as a gatk4 setting rather than just one for both. Do you think it's worth making this split now, or is an update to GATK4 in the SomaticSeq wrapper coming soon?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/bioinform/somaticseq/pull/50#issuecomment-360636928, or mute the thread https://github.com/notifications/unsubscribe-auth/AD3lYGdu3JkTs9zOFbG1rKLAWVG_lBamks5tORAAgaJpZM4RtfIv .

[0xaf1f]

That'd be great. Thanks!

[lethalfang]

Hi Afif,

I've been using GATK3's CombineVariants to merge VCF files. GATK4 doesn't have its equivalent. I could switch out GATK3 CombineVariants, but that'll probably take me a little while to do and then validate. So for now, should I still merge your pull request?

Best,

On Fri, Jan 26, 2018 at 11:33 AM, Afif Elghraoui notifications@github.com wrote:

That'd be great. Thanks!

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/bioinform/somaticseq/pull/50#issuecomment-360882429, or mute the thread https://github.com/notifications/unsubscribe-auth/AD3lYEn8LgWYoDjrn4aStqjZgBBAi9riks5tOihugaJpZM4RtfIv .

[0xaf1f]

51? Well, I guess for now, it would break if anyone tried to run it as written, so maybe it's not a good idea to document it while it's known to be broken.

On the other hand, I saw that you did have support for running SomaticSeq without GATK at all, so maybe we should just not pass --gatk to the invocation of the SomaticSeq wrapper until GATK4 support there is ready. What do you think about that?

[lethalfang]

Yes, we can do that, not having --gatk argument at all. Without --gatk, it'll just cat all the VCF and then sort using this command:

cat $all_snp | egrep -v '^#' | awk -F "\t" '{print $1 "\t" $2 "\t.\t" $4

"\t" $5}' | sort | uniq | awk -F "\t" '{print $1 "\t" $2 "\t" $3 "\t" $4 "\t" $5 "\t" "." "\t" "PASS" "\t" "."}' | cat <(echo -e '##fileformat=VCFv4.1\n#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO') - | $MYDIR/utilities/vcfsorter.pl ${hg_dict} - > ${merged_dir}/CombineVariants_MVJSD.snp.vcf

On Thu, Feb 15, 2018 at 1:08 PM, Afif Elghraoui notifications@github.com wrote:

51 https://github.com/bioinform/somaticseq/pull/51? Well, I guess for

now, it would break if anyone tried to run it as written, so maybe it's not a good idea to document it while it's known to be broken.

On the other hand, I saw that you did have support for running SomaticSeq without GATK at all, so maybe we should just not pass --gatk to the invocation of the SomaticSeq wrapper until GATK4 support there is ready. What do you think about that?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/bioinform/somaticseq/pull/50#issuecomment-366062074, or mute the thread https://github.com/notifications/unsubscribe-auth/AD3lYHHWrRruZwZlatYMfZOyLQx1FcsTks5tVJzEgaJpZM4RtfIv .

[0xaf1f]

Ok. I'll file an issue for this GATK4 thing just for tracking purposes, and then I'll add another commit to #51.