cond.topsnp.meta

#!/bin/bash

## PLEASE ONLY ABSOLUTE PATHS

# this stays constant across studies
ensg=$1
shift
prot=$1
shift
condition=$1
shift
outfile=$1
shift
setfile=$1
shift
indexcohort=$1
shift

DIR="$( cd "$( dirname "${BASH_SOURCE[0]}" )" >/dev/null 2>&1 && pwd )"

# this is study specific
SS=("$@")
names=()
phenos=()
plinkfiles=()
matrices=()
gdss=()
assocs=()
indexcohortindex="none"
i=0
for ssarg in ${SS[@]}; do
	IFS=',' read -r -a csfields <<< "$ssarg"
		names+=( "${csfields[0]}" )
		phenos+=( "${csfields[1]}" )
		plinkfiles+=( "${csfields[2]}" )
		matrices+=( "${csfields[3]}" )
		gdss+=( "${csfields[4]}" )
		assocs+=( "${csfields[5]}" )
		if [[ "${csfields[0]}" == $indexcohort ]]; then
			indexcohortindex=$i
		fi
		i=$(( i + 1 ))
done

if [[ "$indexcohortindex" == "none" ]]; then
	echo Study $indexcohort was not found among ${names[*]}
fi

for i in ${!SS[@]}; do
	echo study $i is named ${names[$i]} #, has pheno ${phenos[$i]}, plink ${plinkfiles[$i]}, matrix ${matrices[$i]}, gds ${gdss[$i]}, assoc ${assocs[$i]}
done
echo index cohort is ${names[$indexcohortindex]}

mkdir $outfile || exit
cd $outfile

$DIR/cond.topsnp.internal $ensg $condition ${phenos[$indexcohortindex]} ${plinkfiles[$indexcohortindex]} ${matrices[$indexcohortindex]} ${gdss[$indexcohortindex]} $setfile ${assocs[$indexcohortindex]} $indexcohort.$prot.$ensg.$condition.topsnp

condsnp=$(head -1 $indexcohort.$prot.$ensg.$condition.topsnp/tocondition.raw | awk '{print $NF}'| sed 's/_.*//')

echo Now conditioning on other cohorts for $condsnp...

for i in ${!SS[@]}; do
	if [[ "${names[$i]}" != "$indexcohort" ]]; then
		echo Processing cohort "${names[$i]}" ...
		$DIR/cond.snplist $ensg $condition ${phenos[$i]} ${plinkfiles[$i]} ${matrices[$i]} ${gdss[$i]} $setfile $condsnp ${names[$i]}.$prot.$ensg.$condition.${condsnp/:/_}
		if [[ ! -f ${names[$i]}.$prot.$ensg.$condition.${condsnp/:/_}/tocondition.raw ]]; then
			echo
			echo WARNING : Variant is absent from cohort "${names[$i]}"
			echo           Analysing without conditional
			echo
			rm -r ${names[$i]}.$prot.$ensg.$condition.${condsnp/:/_}
			$DIR/reanalyse.single.burden $ensg $condition ${phenos[$i]} ${plinkfiles[$i]} ${matrices[$i]} ${gdss[$i]} $setfile $condsnp ${names[$i]}.$prot.$ensg.$condition.${condsnp/:/_}
		fi
	fi
done

echo
echo META ANALYSIS
echo
grep -w '^'$ensg.$condition $setfile>redux.variantset
Rscript -e 'library(GMMAT);library(data.table)
meta=SMMAT.meta(meta.files.prefix=c('$(find -name "*.score.1" | sed 's/.score.1//;s/^/\"/;s/$/\"/'| tr '\n' ',' | sed 's/,$/\n/')'), n.files=rep(1, '"${#names[@]}"'),
        group.file="redux.variantset",
        MAF.range=c(1e-10, 0.05),
        MAF.weights.beta = c(1,1),
        miss.cutoff=0.01,
        tests=c("O", "E"),
        rho=(0:10)/10,
        use.minor.allele=T
          )

fwrite(meta, "'$ensg.$prot'.cond.out", sep="\t", na=NA, quote=F)
print(meta)'