more time efficient computation of contig-correspondance between asse…

…mblies

src/convert_coord/corresp_contigs.sh

@@ -57,49 +57,22 @@ eval "$run_fun" <<CORR
 #BSUB -n 36
 #BSUB -R "span[ptile=36]"
 
-MAX_PROC=32
-master_pid=\$\$
-echo "\$master_pid"
 
-# Used to display loading bar
-source src/misc/jobs_manager.sh
-# Total contigs to process
-tot=\$(grep ">" $OLD_REF | wc -l )
 
 # Get start position of all contigs in the ordered assembly and whether
 # They have been reversed and/or complemented. Store infos in a file.
 
 declare -i n=0
-grep ">" $OLD_REF | while read tig
-do
- ((n++))
-
- # Loading bar animation showing N jobs submitted
- prettyload \$n \$tot
- # ps -eo ppid | grep -w "\$master_pid" | wc -l
- # Keeping track of the number of running child processes
- while [ \$(pgrep -c -P\$master_pid) -ge \$MAX_PROC ]
- do
- # Hanging to limit parallel processes
- sleep 1
- done
-
- # Setting region size longer than whole genome -> resized to contig
- ( tname=\$(echo \$tig | sed -n 's/>\([^ ]*\) .*/\1,/p' | tr -d '\n')
- new_coord=\$(python2 src/convert_coord/contig2chr.py --pos1 "\${tname}0" \
+grep ">" $OLD_REF \
+ | sed -n 's/>\([^ ]*\) .*/\1,0/p' \
+ | python2 src/convert_coord/contig2chr.py \
  --region_size 1000000000 \
+ --include_input \
  $OLD_REF \
- $NEW_REF)
- echo \${tname}\${new_coord} >> $tmp_dir/tmp-\$n.csv ) &
-done
-wait
-
-# Merging all files into a single one
-find $tmp_dir/ -name "*.csv" -type f -maxdepth 1 | \
-xargs cat > $CORRESP_GFF
-
-# Sort lines according to contig names
-sort -k1,1 -o $CORRESP_GFF $CORRESP_GFF
+ $NEW_REF \
+ 2> /dev/null \
+ | cut -d ',' -f2 --complement \
+ > $CORRESP_GFF
 
 # insert header
 sed -i "1s/^/tig,chr,start,region_size,transform\\n/" $CORRESP_GFF