V2

script/2_make_db.sh

@@ -20,11 +20,11 @@ module purge
 
 
 HOME="/home/mcariou/2021_legio/"
-OUT=$HOME"blastdb/phyloref/"
+OUT=$HOME"blastdb/phyloref_V2/"
 CAT=$OUT"/cat_phyloref_cds.fasta"
 CATPROT=$OUT"/cat_phyloref_pep.fasta"
 Trans="/home/mcariou/2020_Attaiech/prot_db/Transdecoder/"
-TAB=$HOME"/phylolegio/doc/tabAss.txt"
+TAB=$HOME"/phylolegio/doc/tabAss_V2.txt"
            
 mkdir -p $OUT
 
@@ -34,14 +34,14 @@ mkdir -p $OUT
 if [[ -s $CAT ]] ; then 
     echo "cat already exists"
 else
-    for genome in `cat $TAB | sed '1d'| awk '{print $1}'`
+    for genome in `cat $TAB | sed '1d'| awk -F ";" '{print $1}'`
     do
     file=${Trans}/${genome}*/longest_orfs.cds
     file2=`echo $file`
         if [[ -s $file2 ]] ; then
         cat $file2 >> $CAT
         else
-        echo "doesn't: $file2"
+        echo "doesn't: $file"
         fi
     done
 fi

script/psmn/runscript_78Lp_V2.sh

+#!/bin/bash
+#$ -S /bin/bash
+## name of the job to follow them
+#$ -N runscript_V2
+## name of the queue to be used
+#$ -q E5-2670deb*,E5-2667v2*,E5-2667v4*
+#$ -cwd
+#$ -V
+## where to put the log files (output and error) automatically generated by the cluster (different from the .log generated by DGINN)
+## the dirs must exist before job is launched
+#$ -o /home/mcariou/2021_legio/log/
+#$ -e /home/mcariou/2021_legio/log/
+module load trimal
+
+/home/mcariou/2021_legio/phylolegio/script/3_make_PSIblast.sh ~/2021_legio/blastdb/phyloref_V2/phyloref_prot ~/2021_legio/genes/78Lp_uniprot.fasta ~/2021_legio/out_blastn/
+
+
+
+/home/mcariou/2021_legio/phylolegio/script/4_parse_PSIblast.sh  ~/2021_legio/out_blastn/78Lp_uniprot_V2.psiblast  ~/2021_legio/phylolegio/doc/tabAss_V2.txt ~/2021_legio/fasta/78Lp_V2 ~/2021_legio/genes/78Lp_uniprot.fasta 0.0001 0.5 0.5 1
+
+
+/home/mcariou/2021_legio/phylolegio/script/5_cat_aln_phy.sh ~/2021_legio/fasta/78Lp_V2/
+
+
+# fin