#!/bin/tcsh # run per genome dir # add formatdb call here? predict exon split? set nb=/bio/bio-grid/ncbi-20060507/ncbi/bin foreach dir (scaffold_*) set faz="$dir/*_caf060210.fa.gz" set fna=`echo $faz | sed -e's/.gz//'` gzcat $faz | $nb/formatdb -pF -n $fna -i stdin #set nsq="$dir/*_caf060210.fa.nsq" #set fna=`echo $nsq | sed -e's/.nsq//'` echo "# megablast start" $fna , `date` foreach predict ($dir/*-exons.fa) echo "# megablast query=$predict" $nb/megablast -W11 -t16 -N0 -gT -e0.05 -D3 -i $predict -d $fna -o $predict.mbout end echo "# megablast done" $fna , `date` end exit # ---- formatdb ------ foreach faz (scaffold_*/*_caf060210.fa.gz) echo "# formatdb $faz" set fna=`echo $faz | sed -e's/.gz//'` gzcat $faz | $nb/formatdb -pF -n $fna -i stdin end #--- exon split ---- foreach faz (scaffold*/*_caf060210_exons.fa.gz) echo "# split $faz by predictors" gzcat $faz | env fin=$faz perl -ne 'use FileHandle;\ if(/^>(\D+)/){ $gr=$1; $gr=~s/_$//; $fh=undef; $skip=0; $skip= ($gr=~m/DGIL_SNO|BATZ_CNA|RGUI_GID|^TRd\w+/); next if($skip)\ $fn=$ENV{fin}; $fn=~s,/[^/]*$,/$gr-exons.fa, or die; \ $fh=$fhs{$fn}; unless($fh){ $fhs{$fn}= $fh= new FileHandle "> $fn"; } }\ print $fh $_ unless($skip);' end