#########################################################################################
#											#
#                Merlin									#
#											#
#########################################################################################


# make a connecting file to link up the families and supply zygosity if required
awk ' { print  $1, $2, $3, $4, $5, "0"}  ' HM3ceu.fam > zyg.ped
# add in the parents
awk ' { print  $1, $3 }' zyg.ped | grep -v "0$" | sort | uniq > temp
awk ' { print  $1, $2 "0 0 1 0"} ' temp >> zyg.ped
awk ' { print  $1, $4 }' zyg.ped | grep -v "0$" | sort | uniq > temp
awk ' { print  $1, $2 "0 0 2 0"} ' temp >> zyg.ped

# make a dat file for zyg.ped
echo 'Z Zygosity' > zyg.dat

### From the gt folder

# run association - score test
merlin -p zyg.ped,HM3.pheno -d zyg.dat,HM3.pheno.dat -m HM3ceu22.map.gz \
 --singlepoint --trim --fastAssoc --tabulate --prefix fastAssoc --pdf

# run association - Likelihood-ratio test
merlin -p zyg.ped,HM3.pheno -d zyg.dat,HM3.pheno.dat -m HM3ceu22.map.gz  --singlepoint \
 --trim --assoc --tabulate --prefix assoc --pdf

### From the imp folder


#converting from mach to merlin-ofline format for hapmap2 imputed data
#requires the legend, mlinfo, mldose and fam files

./mach2merlin.pl -l genotypes_chr22_CEU_r22_nr.b36_fwd_legend.txt \
-i imputed.mlinfo -d imputed.mldose.gz -f HM3ceu.fam -b 10000


# loop over chromosomes
for ((i=1; i<=22; i++))
#for i in 22
do

# loop over parts
for ((j=1; j<=23; j++))
#for j in 4
do

# exclude parts that are not present
if test -f $GR/infer_format_"$i"."$j".dat.gz
then

# run merlin-offline
merlin-offline \
-m         prepared_earlier/infer_format_"$i".map.gz \
-f         prepared_earlier/infer_format_"$i".freq.gz \
--pedinfer prepared_earlier/infer_format_"$i"."$j".ped.gz \
--datinfer prepared_earlier/infer_format_"$i"."$j".skip.gz \
-p         zyg.ped,HM3.pheno \
-d         zyg.dat,HM3.pheno.dat \
--useCovariates --tabulate --prefix chr"$i"_"$j" > chr"$i"_"$j".out

fi
done
done