################################################################
# Running the GAVIN+ variant interpretation tool for diagnostics
################################################################

## Download GAVIN+ 1.3 beta and data bundle
mkdir gavin-tools
cd gavin-tools
wget http://molgenis.org/downloads/gavin/GAVIN-Plus-1.3beta.jar
wget https://github.com/joerivandervelde/gavin-plus/raw/master/src/test/resources/bundle_r1.2/CGD_26jun2018.txt.gz
wget https://github.com/joerivandervelde/gavin-plus/raw/master/src/test/resources/bundle_r1.2/FDR_allGenes_r1.2.tsv
wget https://github.com/joerivandervelde/gavin-plus/raw/master/src/test/resources/bundle_r1.2/GAVIN_calibrations_r0.5.tsv
wget https://github.com/joerivandervelde/gavin-plus/raw/master/src/test/resources/bundle_r1.2/clinvar.vkgl.patho.26june2018.vcf.gz
cd ..

## Download demo files (note that the VCF has been annotated by SnpEff, ExAC, GoNL and CADD-SNVs)
wget http://molgenis.org/downloads/gavin/demo/GAVIN-Plus_Demo_1000G_Spiked.vcf
wget http://molgenis.org/downloads/gavin/demo/GAVIN-Plus_Demo_1000G_Spiked.fromCadd.tsv

## Show all options of GAVIN+, this now includes options to keep all variants and/or sample data
java -jar gavin-tools/GAVIN-Plus-1.3beta.jar -h

## Run analysis on GAVIN-Plus_Demo_1000G_Spiked.vcf
java -jar gavin-tools/GAVIN-Plus-1.3beta.jar -i GAVIN-Plus_Demo_1000G_Spiked.vcf -o GAVIN-Plus_Demo_1000G_Spiked.RVCF.firstpass.vcf -m CREATEFILEFORCADD -c GAVIN-Plus_Demo_1000G_Spiked.toCadd.tsv -p gavin-tools/clinvar.vkgl.patho.26june2018.vcf.gz -d gavin-tools/CGD_26jun2018.txt.gz -f gavin-tools/FDR_allGenes_r1.2.tsv -g gavin-tools/GAVIN_calibrations_r0.5.tsv

## Inspect the results, how many interesting variants did we find?
grep -v "#" GAVIN-Plus_Demo_1000G_Spiked.RVCF.firstpass.vcf | wc -l

## Note that not all variants could be assessed due to missing CADD scores
## these are written out in this file, which can be uploaded and scores
## using the CADD web service (http://cadd.gs.washington.edu/score)
more GAVIN-Plus_Demo_1000G_Spiked.toCadd.tsv

## However - the output is included in the demo files for your convenience.
## Use GAVIN-Plus_Demo_1000G_Spiked.fromCadd.tsv to run the complete analysis:
java -jar gavin-tools/GAVIN-Plus-1.3beta.jar -i GAVIN-Plus_Demo_1000G_Spiked.vcf -o GAVIN-Plus_Demo_1000G_Spiked.RVCF.vcf -m ANALYSIS -c GAVIN-Plus_Demo_1000G_Spiked.fromCadd.tsv -p gavin-tools/clinvar.vkgl.patho.26june2018.vcf.gz -d gavin-tools/CGD_26jun2018.txt.gz -f gavin-tools/FDR_allGenes_r1.2.tsv -g gavin-tools/GAVIN_calibrations_r0.5.tsv

## Inspect the results, how many variants are there now?
grep -v "#" GAVIN-Plus_Demo_1000G_Spiked.RVCF.vcf | wc -l
diff GAVIN-Plus_Demo_1000G_Spiked.RVCF.firstpass.vcf GAVIN-Plus_Demo_1000G_Spiked.RVCF.vcf