add foundation taxonomy option; flake8

* add foundation taxonomy option. This is needed if you are supplying a TREE foundation!

ghosttree/extensions/cluster.py

@@ -42,5 +42,5 @@ def preprocess_extension_tree_sequences(extension_sequences_fp,
 
     """
     similarity_threshold = str(similarity_threshold)
-    os.system("sumaclust -g -f -t "+similarity_threshold+" -O " +
+    os.system("sumaclust -g -f -t " + similarity_threshold + " -O " +
               "" + otu_formatted_fp + " " + extension_sequences_fp + "")

ghosttree/scaffold/hybridtree.py

@@ -18,7 +18,7 @@ def extensions_onto_foundation(otu_file_fh, extension_taxonomy_fh,
                                extension_seq_fh,
                                foundation_fh,
                                ghost_tree_fp,
-                               graft_level):
+                               graft_level, foundation_taxonomy):
     """Combines two genetic databases into one phylogenetic tree.
 
     Some genetic databases provide finer taxonomic resolution,
@@ -103,15 +103,16 @@ def extensions_onto_foundation(otu_file_fh, extension_taxonomy_fh,
         foundation_tree = \
             _make_nr_foundation_newick(foundation_fh,
                                        extension_genus_accession_list_dic,
-                                       graft_letter)
+                                       graft_letter, foundation_taxonomy)
 
     if sniffer_results == 'fasta':
         nr_foundation_alignment = \
             _make_nr_foundation_alignment(foundation_fh,
                                           extension_genus_accession_list_dic,
                                           graft_letter)
         skbio.io.write(nr_foundation_alignment,
-                       into=ghost_tree_fp + "/nr_foundation_alignment_gt.fasta",
+                       into=ghost_tree_fp +
+                       "/nr_foundation_alignment_gt.fasta",
                        format="fasta")
         foundation_tree, all_std_error = \
             _make_foundation_tree(ghost_tree_fp +
@@ -138,8 +139,8 @@ def extensions_onto_foundation(otu_file_fh, extension_taxonomy_fh,
                                    stdout=subprocess.PIPE,
                                    stderr=subprocess.PIPE)
         std_output, std_error = process.communicate()
-        all_std_error += "FastTree warnings for genus "+key_node+" are:\n" \
-                         + str(std_error) + "\n"
+        all_std_error += "FastTree warnings for genus " + key_node + \
+                         " are:\n" + str(std_error) + "\n"
         mini_tree = skbio.io.read("tmp/mini_tree_gt.nwk", format='newick',
                                   into=skbio.TreeNode)
         node.extend(mini_tree.root_at_midpoint().children[:])
@@ -159,7 +160,7 @@ def _make_accession_id_file(ghost_tree_fp):
     output = open(ghost_tree_fp + "/ghost_tree_extension_accession_ids.txt",
                   "w")
     for node in ghosttree.tips():
-        output.write(str(node.name)+"\n")
+        output.write(str(node.name) + "\n")
     output.close()
 
 
@@ -170,7 +171,7 @@ def _make_mini_otu_files(key_node, extension_genus_accession_list_dic, seqs):
 
     for seq in seqs:
         if seq.metadata['id'] in keep:
-            fasta_format = ">"+seq.metadata['id']+"\n"+str(seq)+"\n"
+            fasta_format = ">" + seq.metadata['id'] + "\n" + str(seq) + "\n"
             output_file.write(fasta_format)
     output_file.close()
 
@@ -229,33 +230,31 @@ def _create_taxonomy_dict(extension_taxonomy_fh, graft_level):
 
 def _make_nr_foundation_newick(foundation_fh,
                                extension_genus_accession_list_dic,
-                               graft_letter):
+                               graft_letter, foundation_taxonomy):
     global foundation_accession_genus_dic
     foundation_accession_genus_dic = {}
     all_genus_list = list(extension_genus_accession_list_dic.keys())
     foundation_tree = skbio.io.read(foundation_fh, format='newick',
                                     into=skbio.TreeNode,
                                     convert_underscores=False)
 
-    foundation_taxonomy = 'minitaxonomy_foundation.txt'  # (TODO!!!!)
     foundation_unique_accessions = []
-    with open(foundation_taxonomy, 'U') as fin:
-        for line in fin:
-            splitline = line.split('\t')
-            accession = splitline[0].strip()
-            foundation_taxonomy = splitline[1].strip()
-            for graft_taxa in all_genus_list:
-                if_case = (re.search(";" + graft_taxa.lower() + ";",
-                                     foundation_taxonomy.lower()) or
-                           re.search(graft_letter + "__" + graft_taxa.lower() + ";",
-                                     foundation_taxonomy.lower()) or
-                           re.search(";" + graft_taxa.lower(),
-                                     foundation_taxonomy.lower()))
-
-                if if_case:
-                    all_genus_list.remove(graft_taxa)
-                    foundation_accession_genus_dic[accession] = graft_taxa
-                    foundation_unique_accessions.append(accession)
+    for line in foundation_taxonomy:
+        splitline = line.split('\t')
+        accession = splitline[0].strip()
+        foundation_taxonomy = splitline[1].strip()
+        for graft_taxa in all_genus_list:
+            if_case = (re.search(";" + graft_taxa.lower() + ";",
+                                 foundation_taxonomy.lower()) or
+                       re.search(graft_letter + "__" + graft_taxa.lower() + ";",
+                                 foundation_taxonomy.lower()) or
+                       re.search(";" + graft_taxa.lower(),
+                                 foundation_taxonomy.lower()))
+
+            if if_case:
+                all_genus_list.remove(graft_taxa)
+                foundation_accession_genus_dic[accession] = graft_taxa
+                foundation_unique_accessions.append(accession)
 
     sheared_tree = foundation_tree.shear(foundation_unique_accessions)
     return sheared_tree

scripts/ghost-tree

@@ -179,11 +179,17 @@ def scaffold():
               default='g',
               help="Taxonomic level to graft extensions to foundation; this "
                    "option is experimental; default is g (genus)")
+@click.option('-i', '--foundation-taxonomy', type=click.File('U'))
 @click.option("--stderr", is_flag=True, help="'saves error log file")
-@click.option("--foundation_alignment", is_flag=True, help="saves non-redundant foundation alignment file")
-@click.option("--foundation_tree", is_flag=True, help="saves foundation phylogenetic tree file")
-@click.option("--exclude_id_list", is_flag=True, help="will not save accession id file")
-def hybrid_tree(graft_level, stderr, foundation_alignment, foundation_tree, exclude_id_list,
+@click.option("--foundation-alignment", is_flag=True,
+              help="saves non-redundant foundation alignment file")
+@click.option("--foundation-tree", is_flag=True,
+              help="saves foundation phylogenetic tree file")
+@click.option("--exclude-id-list", is_flag=True,
+              help="will not save accession id file")
+def hybrid_tree(graft_level, foundation_taxonomy, stderr, foundation_alignment,
+                foundation_tree,
+                exclude_id_list,
                 extension_trees_otu_map, extension_trees_taxonomy_file,
                 extension_trees_sequence_file, foundation_file,
                 ghost_tree_output_folder):
@@ -248,7 +254,7 @@ def hybrid_tree(graft_level, stderr, foundation_alignment, foundation_tree, excl
                                               extension_trees_sequence_file,
                                               foundation_file,
                                               ghost_tree_output_folder,
-                                              graft_level)
+                                              graft_level, foundation_taxonomy)
 
     if stderr:
         logfile = open(ghost_tree_output_folder + "/ghost_tree_log.txt", "w")

setup.cfg

@@ -1,4 +1,5 @@
 [flake8]
-filename = *
+filename = *.py
 exclude = *.pyc, __pycache__
 max-line-length = 120
+ignore=E722