build-classify-eMERGE

#!/bin/sh

DISEASE=sitosterolemia
VCFFILE=eMERGE_EMRPhenome/CHR2.recode.vcf.gz

SUFFIX=$1
MAF=$2

GRAPH=eMERGE_${DISEASE}_CHR2_${SUFFIX}

TRAININGFILE=eMERGE_${DISEASE}.labels.txt
TESTINGFILE=eMERGE_${DISEASE}.testing.labels.txt

TRAININGPATHS=eMERGE_${DISEASE}_CHR2_${SUFFIX}.paths.txt
TESTINGPATHS=eMERGE_${DISEASE}_CHR2_${SUFFIX}.testing.paths.txt
BOTHPATHSPREFIX=eMERGE_${DISEASE}_CHR2_${SUFFIX}.both
BOTHPATHS=${BOTHPATHSPREFIX}.paths.txt

## build the graph and training paths
echo "#######################################################"
echo "########## BUILDING GRAPH AND TRAINING PATHS ##########"
echo "#######################################################"
./run-graph-vcf $GRAPH $TRAININGFILE $VCFFILE $MAF
mv eMERGE_${DISEASE}_CHR2_${SUFFIX}.paths.txt.eMERGE_${DISEASE}.labels.txt $TRAININGPATHS

## build the testing paths
echo "############################################"
echo "########## BUILDING TESTING PATHS ##########"
echo "############################################"
./run-graph-vcf-paths $GRAPH $TESTINGFILE $VCFFILE
mv eMERGE_${DISEASE}_CHR2_${SUFFIX}.paths.txt.eMERGE_${DISEASE}.testing.labels.txt $TESTINGPATHS

## classify the graph using training and testing
echo "#################################################"
echo "########## CLASSIFYING GRAPH WITH WEKA ##########"
echo "#################################################"
./classify-graph $GRAPH

## build the combined paths for PRS
cat $TRAININGPATHS $TESTINGPATHS > $BOTHPATHS

## compute PRS on combined paths
echo "#####################################################"
echo "########## COMPUTING PRS ON COMBINED PATHS ##########"
echo "#####################################################"
./run-graphutils-prs $GRAPH $BOTHPATHSPREFIX 0