rename scripts

bin/piggy

@@ -48,6 +48,7 @@ then
 	printf "cd-hit found: %s\n" "$cd_hit_command"
 else
 	printf "Error: cd-hit not found\n"
+	exit
 fi
 
 # mafft
@@ -71,6 +72,7 @@ then
 	printf "mafft found: %s\n" "$mafft_command"
 else
 	printf "Error: mafft not found\n"
+	exit
 fi
 
 # blastn
@@ -94,6 +96,7 @@ then
 	printf "blastn found: %s\n" "$blastn_command"
 else
 	printf "Error: blastn not found\n"
+	exit
 fi
 
 # makeblastdb
@@ -117,6 +120,7 @@ then
 	printf "makeblastdb found: %s\n" "$makeblastdb_command"
 else
 	printf "Error: makeblastdb not found\n"
+	exit
 fi
 
 
@@ -147,7 +151,7 @@ fi
 
 if [ -z "$roary_dir" ]
 then
-	printf "No Roary output directory specified.\n"
+	printf "Error: No Roary output directory specified.\n"
 	exit
 fi
 
@@ -182,7 +186,7 @@ fi
 
 if [ ! -d "$in_dir" ]
 then
-	printf "Input folder doesn't exist.\n"
+	printf "Error: Input folder doesn't exist.\n"
 	exit
 fi
 
@@ -215,7 +219,7 @@ if [ "$isolate_count" -gt 0 ]
 then
 	printf "$isolate_count isolates found.\n"
 else
-	printf "$isolate_count isolates found in the input folder.\n"
+	printf "Error: $isolate_count isolates found in the input folder.\n"
 	exit
 fi	
 
@@ -233,9 +237,9 @@ then
 	mkdir "$out_dir/coordinate_files"
 fi
 
-if [ -f "$out_dir/output_fasta.fasta" ]
+if [ -f "$out_dir/IGR_sequences.fasta" ]
 then
-	rm "$out_dir/output_fasta.fasta"
+	rm "$out_dir/IGR_sequences.fasta"
 fi
 
 if [ ! -d "$out_dir/isolate_intergenic_files" ]
@@ -257,22 +261,20 @@ fi
 
 for isolate in ${isolate_array[@]}
 do	
-	gff_modifier.pl "$out_dir/gff_files/$isolate.gff" >> "$out_dir/log.txt"
+	piggy_gff_modifier.pl "$out_dir/gff_files/$isolate.gff" >> "$out_dir/log.txt"
 
-	gene_intergenic_coordinate_extracter.pl "$isolate" "$out_dir/gff_files/$isolate.gff.modified" "$out_dir/coordinate_files" >> "$out_dir/log.txt"
+	piggy_gene_intergenic_coordinate_extracter.pl "$isolate" "$out_dir/gff_files/$isolate.gff.modified" "$out_dir/coordinate_files" >> "$out_dir/log.txt"
 
 	if [ ! -d "$out_dir/isolate_intergenic_files/$isolate" ]; then
 		mkdir "$out_dir/isolate_intergenic_files/$isolate"
 	fi
 
-	intergenic_sequence_extracter.pl "$isolate" "$out_dir/gff_files/$isolate.gff.modified" "$out_dir/coordinate_files/${isolate}_intergenic_coordinates.tab" "$out_dir" "$out_dir/isolate_intergenic_files/$isolate" >> "$out_dir/log.txt"
+	piggy_intergenic_sequence_extracter.pl "$isolate" "$out_dir/gff_files/$isolate.gff.modified" "$out_dir/coordinate_files/${isolate}_intergenic_coordinates.tab" "$out_dir" "$out_dir/isolate_intergenic_files/$isolate" >> "$out_dir/log.txt"
 done
 
 printf "Clustering IGRs...\n" >> "$out_dir/log.txt";printf "Clustering IGRs...\n"
 
-$cd_hit_command -i $out_dir/output_fasta.fasta -o "$out_dir/output_fasta_clustered.fasta" -T "$threads" -n 10 -d 0 -c "$nuc_identity" -G 0 -s "$len_identity" -aS 0.9 -g 0 -r 1 -mask N -M 0 &> /dev/null
-
-rm "$out_dir/output_fasta_clustered.fasta"
+$cd_hit_command -i $out_dir/IGR_sequences.fasta -o "$out_dir/IGR_sequences_clustered.fasta" -T "$threads" -n 10 -d 0 -c "$nuc_identity" -G 0 -s "$len_identity" -aS 0.9 -g 0 -r 1 -mask N -M 0 &> /dev/null
 
 printf "IGRs clustered.\n" >> "$out_dir/log.txt";printf "IGRs clustered.\n"
 
@@ -291,7 +293,7 @@ then
 	mkdir "$out_dir/cluster_representative_files"
 fi
 
-cluster_file_creator.pl "$out_dir" >> "$out_dir/log.txt"
+piggy_cluster_file_creator.pl "$out_dir" >> "$out_dir/log.txt"
 
 printf "Doing all-vs-all IGR cluster search.\n" >> "$out_dir/log.txt";printf "Doing all-vs-all IGR cluster search.\n"
 
@@ -301,17 +303,17 @@ $blastn_command -db "$out_dir/representative_clusters_db" -query "$out_dir/repre
 
 printf "Merging IGR clusters.\n" >> "$out_dir/log.txt";printf "Merging IGR clusters.\n"
 
-cluster_merger.pl "$out_dir/representative_clusters_blast_out.tab" "$out_dir/representative_clusters_merged.tab" "$out_dir/cluster_intergenic_files" "$out_dir/clusters.txt" "$out_dir/representative_clusters.fasta" "$nuc_identity" "$len_identity"
+piggy_cluster_merger.pl "$out_dir/representative_clusters_blast_out.tab" "$out_dir/representative_clusters_merged.tab" "$out_dir/cluster_intergenic_files" "$out_dir/clusters.txt" "$out_dir/representative_clusters.fasta" "$nuc_identity" "$len_identity"
 
 readarray -t cluster_array < "$out_dir/clusters.txt"
 
 cluster_count=${#cluster_array[@]}
 
 printf "%i IGR clusters found after merging.\n" "$cluster_count" >> "$out_dir/log.txt";printf "%i IGR clusters found after merging.\n" "$cluster_count"
 
-cluster_presence_absence.pl "$out_dir" >> "$out_dir/log.txt"
+piggy_cluster_presence_absence.pl "$out_dir" >> "$out_dir/log.txt"
 
-roary_piggy_combiner.pl "$out_dir" "$roary_dir" >> "$out_dir/log.txt"
+piggy_roary_piggy_combiner.pl "$out_dir" "$roary_dir" >> "$out_dir/log.txt"
 
 if [ "$fast" -eq 0 ]
 then
@@ -329,7 +331,7 @@ do
 		$mafft_command --thread "$threads" --retree 2 --maxiterate 0 --adjustdirection --quiet "$out_dir/cluster_intergenic_files/${cluster}.fasta" 1> "$out_dir/cluster_intergenic_files/${cluster}_aligned_tmp.fasta" 2> /dev/null
 		#$mafft_command --thread "$threads" --localpair --maxiterate 1000 --adjustdirection --quiet "$out_dir/cluster_intergenic_files/${cluster}.fasta" 1> "$out_dir/cluster_intergenic_files/${cluster}_aligned_tmp.fasta" 2> /dev/null
 
-		fasta_converter.pl "$out_dir/cluster_intergenic_files/${cluster}_aligned_tmp.fasta" "$out_dir/cluster_intergenic_files/${cluster}_aligned.fasta"
+		piggy_fasta_converter.pl "$out_dir/cluster_intergenic_files/${cluster}_aligned_tmp.fasta" "$out_dir/cluster_intergenic_files/${cluster}_aligned.fasta"
 
 		cp "$out_dir/cluster_intergenic_files/${cluster}_aligned.fasta" "$out_dir/cluster_intergenic_alignment_files/${cluster}_aligned.fasta"
 	else
@@ -347,24 +349,24 @@ printf "IGR clusters aligned.\n" >> "$out_dir/log.txt";printf "IGR clusters alig
 
 #printf "Calculating gene divergences...\n" >> "$out_dir/log.txt";printf "Calculating gene divergences...\n"
 
-#divergence_checker.pl "$roary_dir/pan_genome_sequences" "$out_dir" "roary_gene_divergences.csv"
+#piggy_divergence_checker.pl "$roary_dir/pan_genome_sequences" "$out_dir" "roary_gene_divergences.csv"
 
 #printf "Gene divergences calculated.\n" >> "$out_dir/log.txt";printf "Gene divergences calculated.\n"
 
 printf "Calculating IGR divergences...\n" >> "$out_dir/log.txt";printf "Calculating IGR divergences...\n"
 
-divergence_checker.pl "$out_dir/cluster_intergenic_alignment_files" "$out_dir" "cluster_IGR_divergences.csv"
+piggy_divergence_checker.pl "$out_dir/cluster_intergenic_alignment_files" "$out_dir" "cluster_IGR_divergences.csv"
 
 printf "IGR divergences calculated.\n" >> "$out_dir/log.txt";printf "IGR divergences calculated.\n"
 
-core_alignment_creator.pl "$out_dir" >> "$out_dir/log.txt"
+piggy_core_alignment_creator.pl "$out_dir" >> "$out_dir/log.txt"
 
 if [ ! -d "$out_dir/switched_region_files" ]
 then
 	mkdir "$out_dir/switched_region_files"
 fi
 
-switched_region_creator.pl "$out_dir/roary_piggy_combined.tab" "$out_dir/cluster_intergenic_files" "$out_dir" >> "$out_dir/log.txt"
+piggy_switched_region_creator.pl "$out_dir/roary_piggy_combined.tab" "$out_dir/cluster_intergenic_files" "$out_dir" >> "$out_dir/log.txt"
 
 if [ ! -d "$out_dir/switched_region_alignment_files" ]
 then
@@ -379,7 +381,7 @@ for switched_region in ${switched_region_array[@]}
 do
 	$blastn_command -query "$out_dir/switched_region_files/${switched_region}.fasta" -subject "$out_dir/switched_region_files/${switched_region}.fasta" -dust no -outfmt 6 -out "$out_dir/blast_out.tab"
 
-	blast_hit=$(blast_parser.pl "$out_dir/blast_out.tab" "$out_dir/switched_region_files/${switched_region}.fasta" "$out_dir/switched_region_files/${switched_region}_aligned.fasta")
+	blast_hit=$(piggy_blast_parser.pl "$out_dir/blast_out.tab" "$out_dir/switched_region_files/${switched_region}.fasta" "$out_dir/switched_region_files/${switched_region}_aligned.fasta")
 
 	rm "$out_dir/blast_out.tab"
 
@@ -388,7 +390,7 @@ do
 		$mafft_command --thread "$threads" --retree 2 --maxiterate 0 --adjustdirection --quiet "$out_dir/switched_region_files/${switched_region}.fasta" 1> "$out_dir/switched_region_files/${switched_region}_aligned_tmp.fasta" 2> /dev/null
 		#$mafft_command --thread "$threads" --localpair --maxiterate 1000 --adjustdirection --quiet "$out_dir/switched_region_files/${switched_region}.fasta" 1> "$out_dir/switched_region_files/${switched_region}_aligned_tmp.fasta" 2> /dev/null
 
-		fasta_converter.pl "$out_dir/switched_region_files/${switched_region}_aligned_tmp.fasta" "$out_dir/switched_region_files/${switched_region}_aligned.fasta"
+		piggy_fasta_converter.pl "$out_dir/switched_region_files/${switched_region}_aligned_tmp.fasta" "$out_dir/switched_region_files/${switched_region}_aligned.fasta"
 	fi
 
 	cp "$out_dir/switched_region_files/${switched_region}_aligned.fasta" "$out_dir/switched_region_alignment_files/${switched_region}_aligned.fasta"
@@ -398,7 +400,7 @@ printf "Candidate switched IGRs aligned.\n" >> "$out_dir/log.txt";printf "Candid
 
 printf "Calculating candidate switched IGR divergences...\n" >> "$out_dir/log.txt";printf "Calculating candidate switched IGR divergences...\n"
 
-divergence_checker.pl "$out_dir/switched_region_alignment_files" "$out_dir" "switched_region_divergences.csv"
+piggy_divergence_checker.pl "$out_dir/switched_region_alignment_files" "$out_dir" "switched_region_divergences.csv"
 
 printf "Candidate switched IGR divergences calculated.\n" >> "$out_dir/log.txt";printf "Candidate switched IGR divergences calculated.\n"
 
@@ -408,7 +410,7 @@ then
 
 	# Needs R, Rscript, ggplot2, reshape2.
 
-	SR_plotter.R "$out_dir" "$nuc_identity" "$len_identity" &> /dev/null
+	piggy_SR_plotter.R "$out_dir" "$nuc_identity" "$len_identity" &> /dev/null
 fi
 
 fi
@@ -419,10 +421,30 @@ then
 
 	# Needs R, Rscript, ggplot2, reshape2.
 
-	R_plotter.R "$out_dir" "$roary_dir" &> /dev/null
+	piggy_R_plotter.R "$out_dir" "$roary_dir" &> /dev/null
 fi
 
 # Cleanup
 
-#rm -r "$out_dir/cluster_intergenic_files"
+mkdir "$out_dir/plots" &> /dev/null
+mv "$out_dir/"*".tif" "$out_dir/plots"
+
+rm -r "$out_dir/isolate_intergenic_files"
+rm -r "$out_dir/cluster_intergenic_files"
+rm -r "$out_dir/cluster_representative_files"
+rm -r "$out_dir/switched_region_files"
+rm -r "$out_dir/isolate_core_IGR_tmp"
+
+rm "$out_dir/clusters.txt"
+rm "$out_dir/isolates.txt"
+rm "$out_dir/switched_regions.txt"
+rm "$out_dir/representative_clusters_db.nhr"
+rm "$out_dir/representative_clusters_db.nin"
+rm "$out_dir/representative_clusters_db.nsq"
+rm "$out_dir/representative_clusters_merged.tab"
+rm "$out_dir/representative_clusters.fasta"
+rm "$out_dir/representative_clusters_blast_out.tab"
+rm "$out_dir/representative_clusters_blast_out_modified.tab"
+rm "$out_dir/IGR_sequences_clustered.fasta"
+rm "$out_dir/IGR_sequences_clustered.fasta.clstr"
 

bin/cluster_file_creator.pl → bin/piggy_cluster_file_creator.pl

@@ -10,7 +10,7 @@
 open OUTPUT_REP, ">$out_dir/representative_clusters.fasta";
 
 $count=0;
-open INPUT, "$out_dir/output_fasta_clustered.fasta.clstr";
+open INPUT, "$out_dir/IGR_sequences_clustered.fasta.clstr";
 while(<INPUT>){
 	$line=$_;
 	chomp $line;
@@ -64,7 +64,7 @@
 close OUTPUT_REP;
 
 $count=0;
-open INPUT, "$out_dir/output_fasta_clustered.fasta.clstr";
+open INPUT, "$out_dir/IGR_sequences_clustered.fasta.clstr";
 while(<INPUT>){
 	$line=$_;
 	chomp $line;

bin/intergenic_sequence_extracter.pl → bin/piggy_intergenic_sequence_extracter.pl

@@ -10,7 +10,7 @@
 $max_len=1000;
 $max_n_prop=0.1;
 
-open OUTPUT, ">>$out_dir/output_fasta.fasta";
+open OUTPUT, ">>$out_dir/IGR_sequences.fasta";
 
 $include=0;
 open INPUT, "$in_file";