How to run GIAB comparisons #7237
Conversation
| continue | ||
|
|
||
| if (len(sys.argv) > 2 and sys.argv[2] == "loose"): |
There was a problem hiding this comment.
I didn't see this option in the readme. When should this be used?
| # extract each of the samples | ||
| INPUT_VCF=gvs.chr20.vcf.gz | ||
|
|
||
| gatk SelectVariants -V ${INPUT_VCF} --sample-name SM-G947Y --select-type-to-exclude NO_VARIATION -O NA12878.gvs.chr20.vcf.gz |
There was a problem hiding this comment.
this may be minor, but can we document explicitly what --select-type-to-exclude NO_VARIATION does? i'm guessing it means only return sites at which this sample has a variant, i.e. exclude ref sites?
|
|
||
| ## script to add "AS_MAX_VQSLOD" to VCFs | ||
| ``` | ||
| for sample in NA12878 SYNDIP BI_HG002 BI_HG003 UW_HG002 |
| tabix warp_tieout_acmg_cohort_v1.chr20.vcf.gz | ||
|
|
||
| INPUT_VCF=warp_tieout_acmg_cohort_v1.chr20.vcf.gz | ||
| SOURCE=warp |
There was a problem hiding this comment.
does this need to be SOURCE="warp" ? and does INPUT_VCF assignment also need quotes? or is this one of those bash things that doesn't matter if there's no spaces?
There was a problem hiding this comment.
right -- only needed for quoting special characters, but I changed it just to be consistent
| @@ -1,5 +1,5 @@ | |||
| PROJECT="spec-ops-aou" | |||
| DATASET="gvs_tieout_acmg_v1" | |||
| DATASET="gvs_tieout_acmg_v2" | |||
| - bcftools | ||
| - tabix | ||
| - python 3.7+ | ||
|
|
There was a problem hiding this comment.
my immediate reaction was that I should pip install them, but that's clearly wrong.
I know I'm a noob with this project, but a little more context on the prereqs would be immensely helfpul and would have saved me a lot of time googling
There was a problem hiding this comment.
conda create --name gvs python=3.8
conda activate gvs
conda install -c bioconda samtools=1.9 --force-reinstall
conda install -c bioconda bcftools
conda install -c bioconda rtg-tools
There was a problem hiding this comment.
also add GATK to prereqs -- should this be done through conda tho? isn't gatk technically part of this repository anyway?
|
|
||
| ## Obtain Truth sample VCFs | ||
|
|
||
| First, create a full cohort extract (as described in README.md) using the desired filter_set_name. Assuming this is in a single gathered VCF of `gvs.vcf.gz` |
There was a problem hiding this comment.
Would this be more like "once you have a full cohort extract you want to compare" ?
| ${BASE_CMD} -b truth/CHM.full.38.vcf.gz -e truth/CHM.gvs.evaluation.bed -c SYNDIP.${SOURCE}.chr20.maxas.vcf.gz -o syndip_${SOURCE}${SUFFIX} | ||
| ``` | ||
|
|
||
| The do the same thing but use all records |
Addresses
https://github.com/broadinstitute/dsp-spec-ops/issues/280
Analysis has been done and delivered, this is primarily documentation of how to do it in the future
If someone wants to test-drive the instructions, there is a GVS VCF at