| 1 |
gpertea |
23 |
#!/bin/tcsh -f |
| 2 |
|
|
|
| 3 |
|
|
|
| 4 |
|
|
set cdb=refseq.mammalian_np.CDS.fa.cidx |
| 5 |
|
|
|
| 6 |
|
|
if ($1"x" == "x") then |
| 7 |
|
|
echo "Usage: prep_sim4.csh fasta_files.." |
| 8 |
|
|
echo "" |
| 9 |
|
|
echo "Must be run from the sim4search subdirectory" |
| 10 |
|
|
echo "and expects a gblat_*.gff and/or gmap_*.gff3 file with gblat scanning results" |
| 11 |
|
|
echo "and $cdb in the current directory to pull RefSeqs for sim4cc" |
| 12 |
|
|
echo "" |
| 13 |
|
|
echo "Example: prep_sim4.csh ../ga_??.fa" |
| 14 |
|
|
exit 1 |
| 15 |
|
|
endif |
| 16 |
|
|
|
| 17 |
|
|
# prepare the name translation table |
| 18 |
|
|
|
| 19 |
|
|
set f=$1; |
| 20 |
|
|
|
| 21 |
|
|
if (! -f $cdb ) then |
| 22 |
|
|
echo "Error: missing file in current directory: $cdb" |
| 23 |
|
|
exit 1 |
| 24 |
|
|
endif |
| 25 |
|
|
|
| 26 |
|
|
set cdir=`pwd` |
| 27 |
|
|
set pdir=$cdir:h |
| 28 |
|
|
set gblat_gff=`ls gblat*.gf{f,f3}` |
| 29 |
|
|
|
| 30 |
|
|
if (! -f $gblat_gff ) then |
| 31 |
|
|
echo "Error: no single gblat*.gf{f,f3} file found in current directory" |
| 32 |
|
|
echo "make sure you're in <organism>/sim4search and the gblat results are there" |
| 33 |
|
|
exit 1 |
| 34 |
|
|
endif |
| 35 |
|
|
set infiles=$gblat_gff |
| 36 |
|
|
|
| 37 |
|
|
set gmap_gff=`ls gmap*.gf{f,f3}` |
| 38 |
|
|
if ( -f $gmap_gff ) then |
| 39 |
|
|
#echo "Error: no single gmap*.gf{f,f3} file found in current directory" |
| 40 |
|
|
#echo "make sure you're in <organism>/sim4search and the gblat/gmap results are there" |
| 41 |
|
|
#exit 1 |
| 42 |
|
|
set infiles="$infiles $gmap_gff" |
| 43 |
|
|
endif |
| 44 |
|
|
|
| 45 |
|
|
set fp=$f:h |
| 46 |
|
|
set ftx="$fp/seqn2fname.lst" |
| 47 |
|
|
set dlst="" |
| 48 |
|
|
set flst="" |
| 49 |
|
|
/bin/rm -f $ftx |
| 50 |
|
|
foreach f ( $* ) |
| 51 |
|
|
set fn=$f:t |
| 52 |
|
|
set bd=$fn:r |
| 53 |
|
|
set sn=`head -1 $f | cut -f1 -d ' ' | cut -b2-` |
| 54 |
|
|
#if (! -d $bd) mkdir $bd |
| 55 |
|
|
echo "$sn\t$bd" >> $ftx |
| 56 |
|
|
#/bin/rm -f $bd/$bd.refseq.gmap.gff3 |
| 57 |
|
|
#/bin/rm -f $bd/$bd.unipr.pmap.gff3 |
| 58 |
|
|
set flst = "$flst $fn" |
| 59 |
|
|
end |
| 60 |
|
|
echo "$fp/seqn2fname.lst created." |
| 61 |
|
|
cat $infiles | trcol.pl -G $ftx | sort -u > anchored.lst |
| 62 |
|
|
set simbat=sim4_grid.cmds |
| 63 |
|
|
/bin/rm -f $simbat |
| 64 |
|
|
gawk '{ print "qsim4cc -p75 -c75 -o wrk_sim4cc.gff3 -q \047"$2"\047"" '$cdir/$cdb' '$pdir/'"$1".fa"}' \ |
| 65 |
|
|
anchored.lst > $simbat |
| 66 |
|
|
|
| 67 |
|
|
echo "$simbat file created, use gridx with it, e.g.:" |
| 68 |
|
|
echo "gridx -q -N -O grdsim4logs -f sim4_grid.cmds -p40 -m gpertea" |
