run-fr-lightgbm

#!/bin/sh

GRAPH=$1
FRPREFIX=$2

MINSUPPORT=$3
MAXP=$4
MINPRI=$5

## make the filtered SVM files
java -server -cp "build/install/pangenomics/lib/*" org.ncgr.pangenomics.genotype.fr.FRUtils -svm -g $GRAPH -p $GRAPH         -f $FRPREFIX \
	--minsupport $MINSUPPORT --maxpvalue $MAXP --minpriority $MINPRI
java -server -cp "build/install/pangenomics/lib/*" org.ncgr.pangenomics.genotype.fr.FRUtils -svm -g $GRAPH -p $GRAPH.testing -f $FRPREFIX \
	--minsupport $MINSUPPORT --maxpvalue $MAXP --minpriority $MINPRI

## remove the extra chunk from the file names
mv $FRPREFIX.$GRAPH.svm.txt $FRPREFIX.svm.txt
mv $FRPREFIX.$GRAPH.testing.svm.txt $FRPREFIX.testing.svm.txt

## fix the LIBSVM format
cut -f1 --complement $FRPREFIX.svm.txt > temp0.train
cut -f1 --complement $FRPREFIX.testing.svm.txt > temp0.test
sed s/ctrl/0/ temp0.train > temp1.train
sed s/ctrl/0/ temp0.test  > temp1.test
sed s/case/1/ temp1.train > pathfrs.train
sed s/case/1/ temp1.test  > pathfrs.test 

## train the classifier
lightgbm config=train.conf

## run predictions, use stats.R to view the results
lightgbm config=predict.conf