added option specify 'none' as the aligner. This allows for alignment…

… to be run separately as requested in #306
gtonkinhill · Sep 24, 2024 · 4dc499e · 4dc499e
1 parent f3d664a
commit 4dc499e
Show file tree

Hide file tree

Showing 4 changed files with 33 additions and 10 deletions.
diff --git a/panaroo/__main__.py b/panaroo/__main__.py
@@ -243,7 +243,7 @@ def get_options(args):
         help=
         "Specify an aligner. Options:'prank', 'clustal', and default: 'mafft'",
         type=str,
-        choices=['prank', 'clustal', 'mafft'],
+        choices=['prank', 'clustal', 'mafft', 'none'],
         default="mafft")
     core.add_argument(
         "--codons",
@@ -550,9 +550,10 @@ def main():
         if args.verbose: print("generating pan genome MSAs...")
         generate_pan_genome_alignment(G, temp_dir, args.output_dir, args.n_cpu,
                                       args.alr, args.codons, isolate_names)
-        core_nodes = get_core_gene_nodes(G, args.core, len(args.input_files))
-        core_names = [G.nodes[x]["name"] for x in core_nodes]
-        concatenate_core_genome_alignments(core_names, args.output_dir, args.hc_threshold)
+        if args.alr!='none':
+            core_nodes = get_core_gene_nodes(G, args.core, len(args.input_files))
+            core_names = [G.nodes[x]["name"] for x in core_nodes]
+            concatenate_core_genome_alignments(core_names, args.output_dir, args.hc_threshold)
     elif args.aln == "core":
         if args.verbose: print("generating core genome MSAs...")
         generate_core_genome_alignment(G, temp_dir, args.output_dir,

diff --git a/panaroo/generate_alignments.py b/panaroo/generate_alignments.py
@@ -40,6 +40,8 @@ def check_aligner_install(aligner):
         command = "prank -help"
     elif aligner == "mafft":
         command = "mafft --help"
+    elif aligner == "none":
+        return True
     else:
         sys.stderr.write("Incorrect aligner specification\n")
         sys.exit()

diff --git a/panaroo/generate_output.py b/panaroo/generate_output.py
@@ -315,6 +315,11 @@ def generate_pan_genome_alignment(G, temp_dir, output_dir, threads, aligner,
                                                        aligner,
                                                        threads)
     else:
+        if aligner=='none':
+            temp_dir = output_dir + "unaligned_gene_sequences/"
+            if not os.path.exists(temp_dir):
+                os.mkdir(temp_dir)
+
         #Multithread writing gene sequences to disk (temp directory) so aligners can find them
         unaligned_sequence_files = Parallel(n_jobs=threads)(
             delayed(output_sequence)(G.nodes[x], isolates, temp_dir, output_dir)
@@ -323,6 +328,10 @@ def generate_pan_genome_alignment(G, temp_dir, output_dir, threads, aligner,
         #remove single sequence files
         unaligned_sequence_files = filter(None, unaligned_sequence_files)
 
+        if aligner=='none':
+            print("No aligner specified. Returning unaligned gene fasta files.")
+            return
+
         #Get Biopython command calls for each output gene sequences
         commands = [
             get_alignment_commands(fastafile, output_dir, aligner, threads)
@@ -531,10 +540,20 @@ def generate_core_genome_alignment(
                                                        output_dir, temp_dir,
                                                        aligner, threads)
     else:
+        if aligner=='none':
+            temp_dir = output_dir + "unaligned_gene_sequences/"
+            if not os.path.exists(temp_dir):
+                os.mkdir(temp_dir)
+
         #Output core node sequences
         unaligned_sequence_files = Parallel(n_jobs=threads)(
             delayed(output_sequence)(G.nodes[x], isolates, temp_dir, output_dir)
             for x in tqdm(core_genes))
+
+        if aligner=='none':
+            print("No aligner specified. Returning unaligned gene fasta files.")
+            return
+
         #remove single sequence files
         unaligned_sequence_files = filter(None, unaligned_sequence_files)
 

diff --git a/panaroo/post_run_alignment_gen.py b/panaroo/post_run_alignment_gen.py
@@ -41,7 +41,7 @@ def get_options():
         help=
         "Specify an aligner. Options:'prank', 'clustal', and default: 'mafft'",
         type=str,
-        choices={'prank', 'clustal', 'mafft'},
+        choices={'prank', 'clustal', 'mafft', 'none'},
         default="mafft")
     core.add_argument(
         "--codons",
@@ -109,11 +109,12 @@ def main():
         if args.verbose: print("generating pan genome MSAs...")
         generate_pan_genome_alignment(G, temp_dir, args.output_dir, args.n_cpu,
                                       args.alr, args.codons, isolate_names)
-
-        core_nodes = get_core_gene_nodes(G, args.core, len(isolate_names))
-        core_names = [G.nodes[x]["name"] for x in core_nodes]
-        concatenate_core_genome_alignments(core_names, args.output_dir, 
-                                           args.hc_threshold)
+
+        if args.alr!='none':
+            core_nodes = get_core_gene_nodes(G, args.core, len(isolate_names))
+            core_names = [G.nodes[x]["name"] for x in core_nodes]
+            concatenate_core_genome_alignments(core_names, args.output_dir, 
+                                            args.hc_threshold)
     elif args.aln == "core":
         if args.verbose: print("generating core genome MSAs...")
         generate_core_genome_alignment(G, temp_dir, args.output_dir,