This applet converts a VCF to a GDS for subsequent use (e.g. STAAR annotations / STAARpipeline). This app does not perform annotations itself - it is merely a data format converter. This applet depends on the SeqVarTools, gdsfmt, and the SeqArray R libraries. To save dependency install time this applet comes with an R library that will be unpacked during runtime and should be visible on the DNAnexus worker.

Clone this github repo to a local directory:

git clone https://github.com/drarwood/vcf2gds

Navigate to a relevant directory within the project directory on the DNAnexus platform to install the applet

dx cd /path/to/install/apps 

Now you are ready to build and upload the applet to the DNAnexus plaform directory:

dx build -f vcf2gds

Navigate to the RAP directory where you want the output to be directed:

dx cd /path/to/where/the/output/should/go

Simply run the applet by specifying the name (and path if required) of the *.vcf.gz input VCF and the filename of the output GDS. Note priority is set to high below which is recommended for long processes to avoid jobs potentially being reset when running as normal priority jobs (see dx run on changing job priority):

dx run /path/to/install/apps/vcf2gds \
  -ivcf_file=/path/to/vcf/file/to/convert/my.vcf.gz \
  -igds_filename=my.gds \
  --priority high \
  -y

The default worker set for this app is mem2_ssd1_v2_x16 with 16 parallel processes. We note that higher memory instance types are required when processing the UKB 200K WGS where all INFO fields have been kept. If you run into memory issues, we suggest either reducing the number of parallel processes used for data conversion or use an instance type with more memory. To set the number of parallel processes, add the -iparallel option as follows:

dx run /path/to/install/apps/vcf2gds \
    -ivcf_file=/path/to/vcf/file/to/convert/my.vcf.gz \
    -igds_filename=my.gds \
    -iparallel=16 \
    --priority high \
    -y