#
# IMPORTANT NOTICE:
#     Assemble will only work with bap version 12.1 or greater
#
# Script to assemble data into a database using a file of file names.
# The batch of readings is screened against two vectors. Any matching
# the vectors have their names added to growing files of vector failures.
# Those that pass are assembled. Those that fail assembly have their names
# added to a growing list of failures. Output from 
# the initial assembly is directed to a file that is deleted, hence
# nothing appears on the screen. The failed files are then recompared
# with the output directed to another file. This file is printed then
# both output files are deleted. The procedure starts by making a copy
# of the database to copy X (preceded by its deletion!!!!).
#
# use: assemble fileoffilenames
#
unset noclobber
onintr bailout
set PROJECT = F02A9
set VECTOR_SCORE = 20
set ASSEMBLY_SCORE = 20
set MAX_READ_PADS = 25
set MAX_CONTIG_PADS = 25
set MAX_MISMATCH_1 = 8
set MAX_MISMATCH_2 = 15
if ($#argv != 1) then
echo "Usage: assemble fileoffilenames"
	exit 1
endif
if  (! -f $1) then
echo "File of file names not found"
	exit 2
endif
if  (! -f `head -1 $1`) then
echo "First file in file of file names not found"
	exit 3
endif
echo -n "Assembling data from file of filenames "; echo -n $1
echo  -n " into data base "; echo $PROJECT
echo -n "Deleting database backup: "; echo -n $PROJECT
echo " version X"
\rm -f $PROJECT.RLX
\rm -f $PROJECT.SQX
\rm -f $PROJECT.ARX
\rm -f $PROJECT.CCX
\rm -f $PROJECT.TGX
bap <<endofinput
3
y
$PROJECT
0

7
1
out$$

24
X
1000

18
y
$1
passm13$$
M13MP18_VECTOR
$VECTOR_SCORE

18
y
passm13$$
passblue$$
BLUE_VECTOR
$VECTOR_SCORE

20
Y
Y
Y
passblue$$
fail$$
1
Y
$ASSEMBLY_SCORE
3
$MAX_READ_PADS
$MAX_CONTIG_PADS
$MAX_MISMATCH_1

7
y

7
1
show$$

20
y
n
y
fail$$
fail2$$
1
y
$ASSEMBLY_SCORE
3
$MAX_READ_PADS
$MAX_CONTIG_PADS
$MAX_MISMATCH_2

25
!

7
y

!
endofinput
cat fail2$$ >> $PROJECT.FAILS
echo -n "Assembly for project ">>show1$$
echo -n  $PROJECT >>show1$$
echo -n " on " >>show1$$
date >> show1$$
echo -n "In directory " >> show1$$
pwd >> show1$$
echo -n "Input file of file names ">>show1$$
echo $1 >> show1$$
echo "Parameters used" >>show1$$
echo -n "Minimum initial match  " >>show1$$
echo $ASSEMBLY_SCORE >> show1$$
echo -n "Maximum pads in each reading  " >>show1$$
echo $MAX_READ_PADS >> show1$$
echo -n "Maximum pads in contig  " >>show1$$
echo $MAX_CONTIG_PADS >> show1$$
echo -n "Maximum percent mismatch (for assembly)  " >>show1$$
echo $MAX_MISMATCH_1 >> show1$$
echo -n "Maximum percent mismatch (for report)  " >>show1$$
echo $MAX_MISMATCH_2 >> show1$$
echo "List of m13mp18 failures">>show1$$
cat $1 | tr -d " " | sort >! sortin$$
cat passm13$$ | tr -d " " | sort >! sortm13$$
comm -23 sortin$$ sortm13$$ >> show1$$
echo "List of bluescribe failures">>show1$$
#assuming already have sortm13$$
cat passblue$$ | tr -d " " | sort >! sortblue$$
comm -23 sortm13$$ sortblue$$ >> show1$$
comm -23 sortin$$ sortm13$$ >> $PROJECT.M13
comm -23 sortm13$$ sortblue$$ >> $PROJECT.BLUE
echo "List of entry failures ">>show1$$
cat fail2$$ >> show1$$
#cp fail$$ show1$$
echo "Alignments for entry failures ">>show1$$
cat show$$ >> show1$$
lpr show1$$
bailout:
\rm fail$$
\rm fail2$$
\rm -f out$$
\rm -f show$$
\rm -f show1$$
\rm -f passm13$$
\rm -f passblue$$
\rm -f sortin$$
\rm -f sortm13$$
\rm -f sortblue$$
#cat $PROJECT.FAILS
;