commit

script/4_parse_PSIblast.sh

@@ -20,9 +20,12 @@ echo "--------------------------------------------------"
 ##################################################################################################################
 module load R
 
-
+echo "TMP version: blastdb from which sequences are retreived and taxo db used by Taxonomizer R package are written in the script, should be parameters"
 BLASTDBNUC="~/2021_legio/blastdb/phyloref/phyloref_nuc" 
-TAXO="/home/mcariou/2021_legio/Taxo/accessionTaxa.sql
+TAXO="/home/mcariou/2021_legio/Taxo/accessionTaxa.sql"
+echo "~/2021_legio/blastdb/phyloref/phyloref_nuc" 
+echo "/home/mcariou/2021_legio/Taxo/accessionTaxa.sql"
+
 
 # variable
 OUT_BLAST=$1
@@ -32,9 +35,7 @@ BLAST_REP=`echo $1 | awk -F "/" '{$NF=""; print $0}' | sed 's/ /\//g'`
 FASTA_REP=$3
 mkdir -p $FASTA_REP
 
-#GENES=`grep ">" $4 | sed 's/>//g'`
-GENES="lpg2300"
-
+GENES=`grep ">" $4 | sed 's/>//g'`
 
 EVAL=$5
 PERCID=$6
@@ -65,31 +66,11 @@ nrow=`cat $subblast | wc -l`
     echo "sequences already retrieved"
     else
     echo "retrieve fasta"
-    Rscript --vanilla $rscript $subblast $EVAL $PERCID $PERCOVERLAP $NPERTAX $FASTA> $FASTA_REP/$Gene/paste_blast_test.log
+    #Rscript --vanilla $rscript $subblast $EVAL $PERCID $PERCOVERLAP $NPERTAX $FASTA> $FASTA_REP/$Gene/paste_blast_test.log
+    Rscript --vanilla $rscript $subblast $EVAL $PERCID $PERCOVERLAP $NPERTAX $FASTA $BLASTDBNUC $TAXO > $FASTA_REP/$Gene/paste_blast_test.log
     fi ;
 
 
-#    if [[ -s $subblast.mod ]] ; then
-#    echo "sp names already retrieved"
-#    else
-#    echo "debut de la boucle:"
-#        for id in `seq $nrow`
-#        do
-#        line=`sed -n ${id}p $subblast`
-#        idncbi=`sed -n ${id}p $subblast | awk '{print $3}' | awk -F[\.] '{print $1}'`
-#        tax=`efetch -db nuccore -id $idncbi -format docsum | xtract -pattern DocumentSummary -element Organism | sed 's/ /_/g'`
-#        echo $tax
-#        echo $line " " $tax >> $subblast.mod
-#        done
-#    fi ;
-
-
-    #if [[ -s $FASTA ]] ; then
-    #echo "sequences already retrieved"
-    #else
-    #echo "retrieve fasta"
-    #Rscript --vanilla $rscript $subblast.mod $EVAL $PERCID $PERCOVERLAP $NPERTAX $FASTA> $FASTA_REP/$Gene/paste_blast.log
-    #fi ;
 
 done
 

script/runscript_78Lp_step34.sh

@@ -12,11 +12,11 @@
 #$ -e /home/mcariou/2021_legio/log/
 
 
-/home/mcariou/2021_legio/phylolegio/script/3_make_PSIblast.sh ~/2021_legio/blastdb/phyloref/phyloref_prot ~/2021_legio/genes/78Lp_uniprot.fasta ~/2021_legio/out_blastn/
+#/home/mcariou/2021_legio/phylolegio/script/3_make_PSIblast.sh ~/2021_legio/blastdb/phyloref/phyloref_prot ~/2021_legio/genes/78Lp_uniprot.fasta ~/2021_legio/out_blastn/
 
 
-## Doesn't accomodate multi query yet
-#/home/mcariou/2021_legio/phylolegio/script/4_parse_PSIblastTMP.sh  ~/2021_legio/out_blastn/78Lp_uniprot.psiblast  ~/2021_legio/phylolegio/doc/tabAss.txt ~/2021_legio/fasta/lpg2300_50_50 ~/2021_legio/genes/lpg2300.fasta 0.0001 0.5 0.5 1
+## IN TRIAL
+/home/mcariou/2021_legio/phylolegio/script/4_parse_PSIblast.sh  ~/2021_legio/out_blastn/78Lp_uniprot.psiblast  ~/2021_legio/phylolegio/doc/tabAss.txt ~/2021_legio/fasta/lpg2300_50_50 ~/2021_legio/genes/78Lp_uniprot.fasta 0.0001 0.5 0.5 1
 
 
 # fin