#
# mergeped 20021213 David L Duffy
#
# Read one or more sib-pair pedigree files and corresponding scripts
# and merge these into a single pedigree file and script
#
#
# > mergeped script1 script2
#
#  script1             script2        -->     mergeped.in
#  ==========          ===========            ===========
#  set loc a qua       set loc a qua          set loc a qua
#  read ped pedfile1   read ped pedfile2      set loc a_v2 qua
#                                             read pedigree mergeped.ped
#
#  pedfile1            pedfile2       -->     mergeped.ped
#  ===========         ===========            ============
#  1 1 x x x 1         1 1 x x m 1            1 1 x x m 1 1 
#  1 2 x x f 2         1 2 x x 2 2            1 2 x x f 2 2 
#  1 3 1 2 m 2         1 3 1 3 m 2            1 3 1 2 m 2 2 
#                      1 4 1 3 m 2            1 4 1 3 m x 2 
#  mergeped.err
#  ============
#  pedfile2: Mothers of 1-3 (2 and 3) do not match!
#
# Model:
# Most individuals are present in more than one pedigree file, but allowed
# to be absent in all but one.
#
# Extra records with the same pedigree and individual ID are deleted
#
# Where pedigree information about an individual differs between
# pedigree files, missing data (ie parents or sex in master record)
# is updated, but incongruent data (ie different parents or sex)
# is flagged.
#
# Loci may present in more than one pedigree files, 
# and loci of the same name in different scripts are assumed to be
# different loci, and will be appropriately relabelled.  
#
# Loci in "mergeped.in" are ordered according to their appearance in the 
# original scripts.
#
AWK=/usr/bin/gawk
ERRLOG="mergeped.err"
SCRIPT="mergeped.in"
PEDFILE="mergeped.ped"
PEDLIST=peds$$
IDLIST=idlist$$
MASTER=0

HELP=0
if [ $# = 0 ] || [ $# = 1 ] 
then
  HELP=1
else
  for i in $*
  do
    if [ "$i" = "-m" ] || [ "$i" = "--master" ]
    then
      shift 1
      MASTER=1
    elif [ ! -f $i ]
    then
      HELP=1
      echo "ERROR: $i does not exist"
      echo
    fi
  done 
fi

if [ $HELP = 1 ]
then
	echo "Usage: mergeped [-m|--master] <script1> <script2> <script3>..."
        echo "Merge sib-pair type pedigree files and scripts"
        echo ""
        echo "eg mergeped script1 script2"
        echo ""
        echo "script1             script2        -->  mergeped.in"
        echo "==========          ===========         ==========="
        echo "set loc a qua       set loc a qua       set loc a qua"
        echo "read ped pedfile1   read ped pedfile2   set loc a_v2 qua"
        echo "                                        read pedigree mergeped.ped"
        echo ""
        echo "pedfile1            pedfile2       -->  mergeped.ped"
        echo "===========         ===========         ============"
        echo "1 1 x x x 1         1 1 x x m 1         1 1 x x m 1 1 "
        echo "1 2 x x f 2         1 2 x x 2 2         1 2 x x f 2 2 "
        echo "1 3 1 2 m 2         1 3 1 3 m 2         1 3 1 2 m 2 2 "
        echo "                    1 4 1 3 m 2         1 4 1 3 m x 2 "
        echo "mergeped.err"
        echo "============"
        echo "pedfile2: Mothers of 1-3 (2 and 3) do not match!"
        exit
fi

#
# Merge the scripts
#
$AWK -v pedfile=$PEDFILE -v pedlist=$PEDLIST '
BEGIN { thisfile=""
}
FILENAME!=thisfile {
        thisfile=FILENAME
        run=0
}
substr($1,1,1) ~ /[!#]/ {
        print $0
}
$1=="run" {
        run=1
}
$1=="set" && substr($2,1,3)=="loc"  && run==0 {
        used[$3]++
        ver=""
        if (used[$3]!=1) ver="_v" used[$3]
        $3=$3 ver
        print $0
}
substr($1,1,3)=="rea" && (substr($2,1,3)=="ped" || substr($2,1,3)=="lin") {
        print $3 > pedlist
}
END   { print "read pedigree " pedfile
        print "run"
        print " "
}' $* > $SCRIPT
#
# Merge the pedigree files
#
if [ $MASTER = 1 ]
then
  head -1 $PEDLIST > $IDLIST
else
  cp $PEDLIST $IDLIST
fi

> $PEDFILE
for fil in `cat $IDLIST`
do
   $AWK '$1 !~ "[#!]" {
                print $1 "_sort_" $2, $1, $2, $3, $4, $5
   }' $fil >> $PEDFILE
done
sort -u -k1,1 $PEDFILE > tmp.ped$$
mv tmp.ped$$ $PEDFILE
#
# Update pedigree structure for each record using other pedigree files
#
> $ERRLOG
for fil in `cat $PEDLIST`
do
   $AWK '$1 !~ "[#!]" { 
                printf "%s%s%s %s %s %s\n", $1, "_sort_", $2, $3, $4, $5
   }' $fil > tmp.ped$$
   sort -u -k1,1 tmp.ped$$ > part2.ped$$
   NMATCH=`join $PEDFILE part2.ped$$ | wc -l`     
   if [ $NMATCH = 0 ]
   then
     echo "No IDs from \"$fil\" match master list!" > $ERRLOG
   else
     join -a1 $PEDFILE part2.ped$$ > tmp.ped$$
     pad tmp.ped$$ | $AWK -v fil=$fil -v errlog=$ERRLOG '
     BEGIN { sex[1]="m"; sex["M"]="m"; sex["m"]="m"
             sex[2]="f"; sex["F"]="f"; sex["f"]="f"
     }
           { if (($4=="x" || $4==".") && $7!="x" && $7!=".") $4=$7
             if (($5=="x" || $5==".") && $8!="x" && $8!=".") $5=$8
             if (($6=="x" || $6==".") && $9!="x" && $9!=".") $6=sex[$9]
    
             if ($4!="x" && $4!="." && $7!="x" && $7!="." && $4!=$7) {
               print fil ": Fathers of " $2 "-" $3 \
                     " (" $4 " and " $7 ") do not match master!" > errlog
             }
             if ($5!="x" && $5!="." && $8!="x" && $8!="." && $5!=$8) {
               print fil ": Mothers of " $2 "-" $3 \
                     " (" $5 " and " $8 ") do not match master!" > errlog
             }
             if ($6!="x" && $6!="." && $9!="x" && $9!="." && sex[$6]!=sex[$9]) {
               print fil ": Sexes of " $2 "-" $3 \
                     " (" $6 " and " $9 ") do not match master!" > errlog
             }
             print $1,$2,$3,$4,$5,$6
     }' > $PEDFILE
   fi
done
cat $ERRLOG
#
# Add phenotypes to each record
#
for fil in `cat $PEDLIST`
do
            
   $AWK '
   BEGIN { FS="[ /\t]+"
   }
         { gsub("^[ ]+","")
           printf "%s%s%s", $1, "_sort_", $2
           for(i=6;i<=NF;i++) {
             printf " %s",$i
           }
           printf "\n"
   }' $fil > tmp.ped$$
   sort -u -k1,1 tmp.ped$$ > part2.ped$$
   join -a1 $PEDFILE part2.ped$$ > tmp.ped$$
   pad tmp.ped$$ > $PEDFILE
done
$AWK '{ $1=""; print $0}' $PEDFILE | bester > tmp.ped$$
mv tmp.ped$$ $PEDFILE
rm part2.ped$$ $IDLIST $PEDLIST
if ! [ -s $ERRLOG ]
then
  rm $ERRLOG
fi