added sp version of pyfasta to bootstrap: updated find_rna: added sto…

…ps to run.sh
gturco · Oct 7, 2013 · 0e1ca1c · 0e1ca1c
1 parent 140cf22
commit 0e1ca1c
Show file tree

Hide file tree

Showing 3 changed files with 106 additions and 11 deletions.
diff --git a/bootstrap.py b/bootstrap.py
@@ -17,7 +17,7 @@ def pip_install(dir_name):
     if processing != 0: raise Exception('can not download processing READ REQUIREMENTS and TRY again!')
     shapely = subprocess.call(['{0}/bin/pip'.format(dir_name), 'install', 'shapely'])
     if shapely != 0: raise Exception('can not download shapely READ REQUIREMENTS and TRY again!')
-    pyfasta = subprocess.call(['{0}/bin/pip'.format(dir_name), 'install', 'pyfasta'])
+    pyfasta = subprocess.call(['{0}/bin/pip'.format(dir_name), 'install','-Iv', 'pyfasta==0.4.5'])
     if pyfasta != 0: raise Exception('can not download pyfasta READ REQUIREMENTS and TRY again!')
     scipy = subprocess.call(['{0}/bin/pip'.format(dir_name), 'install', 'scipy'])
     if scipy != 0: raise Exception('can not download scipy READ REQUIREMENTS and TRY again!')

diff --git a/pipeline/run.sh b/pipeline/run.sh
@@ -26,8 +26,22 @@ DIR=data/${ORGA}_${ORGB}/
 #
 echo checking for unannotated proteins......
 sh coann/co-anno.sh ${ORGA} ${ORGB} $QUOTA $BLAST_DIR
+
+status=$?
+if [ $status -ne 0 ]; then
+  echo "Error running checking for unannotated proteins"
+  exit
+fi
+
 echo finding syntenic regions...
 sh quota.sh $DIR/${ORGA} $DIR/${ORGB} $QUOTA $NCPU
+status=$?
+if [ $status -ne 0 ]; then
+  echo "Error running quota-algn"
+  exit
+fi
+
+
 echo finding cns...
 python scripts/find_cns.py \
 	-q $DIR/${ORGA}.fasta --qbed $DIR/${ORGA}.all.bed \
@@ -41,6 +55,13 @@ python scripts/find_cns.py \
         --pair_fmt pair > $DIR/${ORGA}_${ORGB}.cns.txt 
         ### if using blast+  --blast_path ${BLAST_DIR}\legacy_blast.pl \
 
+status=$?
+if [ $status -ne 0 ]; then
+  echo "Error runnining find cns"
+  exit
+fi
+
+
 python scripts/localdup.py \
        -q $DIR/${ORGA}.fasta --qbed $DIR/${ORGA}.all.bed \
        -s $DIR/${ORGB}.fasta --sbed $DIR/${ORGB}.all.bed \
@@ -55,6 +76,12 @@ python scripts/localdup.py \
        --qdups $DIR/${ORGA}.all.localdups \
        --sdups $DIR/${ORGB}.all.localdups
         ### if using blast+  --blast_path ${BLAST_DIR}\legacy_blast.pl \
+status=$?
+if [ $status -ne 0 ]; then
+  echo "Error runnining finding localdups"
+  exit
+fi
+
 
 python scripts/post_processing/cns_to_fasta.py \
                 -c $DIR/${ORGA}_${ORGB}.cns.txt.local \
@@ -65,23 +92,60 @@ python scripts/post_processing/cns_to_fasta.py \
                 --min_len=18 \
                 > $DIR/${ORGA}_${ORGB}.cns_test.fasta
 
+status=$?
+if [ $status -ne 0 ]; then
+  echo "Error runnining cns2fasta"
+  exit
+fi
+
+
 echo removing cns that have hits in arabidopsis as rna or protein
-wget -O data/at_protein.fasta ftp://ftp.arabidopsis.org/home/tair/Sequences/blast_datasets/TAIR10_blastsets/TAIR10_pep_20101214_updated
-wget -O data/os_protein.fasta ftp://ftp.plantbiology.msu.edu/pub/data/Eukaryotic_Projects/o_sativa/annotation_dbs/pseudomolecules/version_6.1/all.dir/all.pep
+
+if [ ! -f data/at_protein.fasta ]; then
+	wget -O data/at_protein.fasta ftp://ftp.arabidopsis.org/home/tair/Sequences/blast_datasets/TAIR10_blastsets/TAIR10_pep_20101214_updated
+fi
+
+if [ ! -f data/os_protein.fasta ]; then
+	wget -O data/os_protein.fasta ftp://ftp.plantbiology.msu.edu/pub/data/Eukaryotic_Projects/o_sativa/annotation_dbs/pseudomolecules/version_6.1/all.dir/all.pep
+fi
+
 bblast.py -b ${BLAST_DIR}/blastall -p blastx -d data/at_protein.fasta -i $DIR/${ORGA}_${ORGB}.cns_test.fasta -e 0.01 -m 8 -a ${NCPU} -o $DIR/at_protein.blast
 bblast.py -b ${BLAST_DIR}/blastall -p blastx -d data/os_protein.fasta -i $DIR/${ORGA}_${ORGB}.cns_test.fasta -e 0.01 -m 8 -a ${NCPU} -o $DIR/os_protein.blast
 
+status=$?
+if [ $status -ne 0 ]; then
+  echo "Error running bblast"
+  exit
+fi
+
+
 python scripts/post_processing/find_exons.py \
                  --cns $DIR/${ORGA}_${ORGB}.cns.txt.local \
                  -o $DIR \
                  $DIR/at_protein.blast $DIR/os_protein.blast
 
-###NEED TO EDIT find_rna.py SO IT looks for the correct cns fasta file
-wget -O data/thaliana_v10.gff ftp://ftp.arabidopsis.org/home/tair/Genes/TAIR10_genome_release/TAIR10_gff3/TAIR10_GFF3_genes_transposons.gff
-wget -O data/thaliana_v10.description ftp://ftp.arabidopsis.org/home/tair/Genes/TAIR10_genome_release/TAIR10_functional_descriptions
-wget -O data/thaliana_v10.fasta ftp://ftp.arabidopsis.org/home/tair/Genes/TAIR10_genome_release/TAIR10_chromosome_files/TAIR10_chr_all.fas
-perl -pi -e "s/>(.).*/>\$1/" data/thaliana_v10.fasta
-perl -pi -e "s/.{3}//" data/thaliana_v10.gff
+status=$?
+if [ $status -ne 0 ]; then
+  echo "Error running find exons"
+  exit
+fi
+
+
+
+if [ ! -f data/thaliana_v10.gff ]; then
+	wget -O data/thaliana_v10.gff ftp://ftp.arabidopsis.org/home/tair/Genes/TAIR10_genome_release/TAIR10_gff3/TAIR10_GFF3_genes_transposons.gff
+	perl -pi -e "s/.{3}//" data/thaliana_v10.gff
+fi
+
+if [ ! -f data/thaliana_v10.description ]; then
+	wget -O data/thaliana_v10.description ftp://ftp.arabidopsis.org/home/tair/Genes/TAIR10_genome_release/TAIR10_functional_descriptions
+fi
+
+if [ ! -f data/thaliana_v10.fasta ]; then
+	wget -O data/thaliana_v10.fasta ftp://ftp.arabidopsis.org/home/tair/Genes/TAIR10_genome_release/TAIR10_chromosome_files/TAIR10_chr_all.fas
+	perl -pi -e "s/>(.).*/>\$1/" data/thaliana_v10.fasta
+fi
+
 
 python scripts/post_processing/find_rna.py \
 	    --blastpath $BLAST_DIR/blastall \
@@ -93,6 +157,14 @@ python scripts/post_processing/find_rna.py \
    	 -o $DIR \
    	 -d data/thaliana_v10.description
 
+
+status=$?
+if [ $status -ne 0 ]; then
+  echo "Error running find rna"
+  exit
+fi
+
+
 cp $DIR/${ORGA}_${ORGB}.raw.filtered.local  $DIR/${ORGA}_${ORGB}.raw2.filtered.local
 
 python scripts/post_processing/shuffle_protein_cns.py \
@@ -104,6 +176,12 @@ python scripts/post_processing/shuffle_protein_cns.py \
    --pairs $DIR/${ORGA}_${ORGB}.pairs.txt.local \ 
 #creates: $DIR/${ORGA}_${ORGB}.quota.with_new.orthology
 
+status=$?
+if [ $status -ne 0 ]; then
+  echo "Error running shuffle proteins"
+  exit
+fi
+
 
 
 python scripts/post_processing/assign.py \
@@ -118,6 +196,14 @@ python scripts/post_processing/assign.py \
      --pair_fmt pair > $DIR/${ORGA}_${ORGB}.cns.assigned.csv.local
 
 
+status=$?
+if [ $status -ne 0 ]; then
+  echo "Error running assing cns to genes"
+  exit
+fi
+
+
+
 python scripts/post_processing/cns_to_fasta.py \
                 -c $DIR/${ORGA}_${ORGB}.cns.txt.real.local \
                 --qfasta $DIR/${ORGA}.genomic.masked.fasta \
@@ -126,6 +212,15 @@ python scripts/post_processing/cns_to_fasta.py \
                 --sorg ${ORGB} \
                 > $DIR/${ORGA}_${ORGB}.cns.fasta
 
+
+status=$?
+if [ $status -ne 0 ]; then
+  echo "Error running cns to fasta"
+  exit
+fi
+
+
+
 ### creates genelist and cnslist
 sh list.sh ${ORGA} ${ORGB} $QUOTA $QDSGID $SDSGID
 
diff --git a/pipeline/scripts/post_processing/find_rna.py b/pipeline/scripts/post_processing/find_rna.py
@@ -130,10 +130,10 @@ def make_cns_to_at_map(des_file, blast_file, gff, query, subject, outdir):
     #print >>sys.stderr, "skipping main for testing"
     main(options.gff, options.outdir, options.fasta)
 
-    import commands
+    import os
     cmd = "bblast.py -b {0} -p blastn -e 0.001 -m 8 -W 7 -a 6 -i {1}/{2}_{3}.cns_test.fasta -d {1}/at_no_cds.fasta -o {4}".format(options.blastpath,options.outdir, options.query, options.subject, options.blast)
     print >>sys.stderr, "executing\n %s" % cmd
-    print commands.getoutput(cmd)
+    print os.system(cmd)
 
 
     make_cns_to_at_map(options.desc, options.blast, options.gff, options.query,