docs/help; flake8; check for foundation taxonomy

If foundation tree is provided, check for foundation taxonomy. If not there, exit.

Lots of flake8 line length issues (pep8)

ghosttree/filter.py

@@ -26,8 +26,10 @@ def filter_positions(alignment_fh, maximum_gap_frequency,
 def _filter_gap_positions(aln, maximum_gap_frequency):
 
     aln_gap_frequencies = (aln.gap_frequencies(axis='sequence',
-                                               relative=False) / aln._seqs.count())
-    aln_gap_frequencies_boolean = (aln_gap_frequencies <= maximum_gap_frequency)
+                                               relative=False) /
+                           aln._seqs.count())
+    aln_gap_frequencies_boolean = (aln_gap_frequencies <=
+                                   maximum_gap_frequency)
     aln = aln.iloc[:, aln_gap_frequencies_boolean]
 
     return aln

ghosttree/scaffold/hybridtree.py

@@ -9,6 +9,7 @@
 import os
 import shutil
 import subprocess
+import sys
 
 import skbio
 import pandas as pd
@@ -57,10 +58,16 @@ def extensions_onto_foundation(otu_file_fh, extension_taxonomy_fh,
         accession numbers are the same as in the otu_file_fh and
         extension_taxonomy_fh.
 
-    foundation_alignment_fh : filehandle
-        File containing pre-aligned sequences from a genetic marker database
-        in .fasta format. This file refers to the "foundation" of the
-        ghost-tree. Contains accession numbers and taxonomy labels.
+    foundation_fh : filehandle
+        File containing EITHER pre-aligned sequences from a genetic marker
+        database in .fasta format OR a newick tree. This file refers to the
+        "foundation" of the ghost-tree.
+
+        .fasta contains accession numbers *and* taxonomy labels.
+
+        .nwk tree is a tree with accession numbers. MUST supply a foundation
+        taxonomy if using a tree as a foundation. Pass this via the
+        --foundation-taxonomy option. see --help for details
 
     ghost_tree_fp : folder
         Output folder contains files including:
@@ -100,6 +107,10 @@ def extensions_onto_foundation(otu_file_fh, extension_taxonomy_fh,
     sniffer_results = skbio.io.sniff(foundation_fh)[0]
 
     if sniffer_results == 'newick':
+        if foundation_taxonomy is None:
+            sys.exit("ghost-tree error: You must provide a foundation "
+                     "taxonomy if using a foundation tree. Pass the taxonomy "
+                     "file using the '--foundation-taxonomy' flag.")
         foundation_tree = \
             _make_nr_foundation_newick(foundation_fh,
                                        extension_genus_accession_list_dic,
@@ -144,8 +155,8 @@ def extensions_onto_foundation(otu_file_fh, extension_taxonomy_fh,
         mini_tree = skbio.io.read("tmp/mini_tree_gt.nwk", format='newick',
                                   into=skbio.TreeNode)
         node.extend(mini_tree.root_at_midpoint().children[:])
-    print('GRAFT LEVEL: ', graft_letter)
-    print(foundation_tree.ascii_art())
+    # print('GRAFT LEVEL: ', graft_letter)
+    # print(foundation_tree.ascii_art())
     shutil.rmtree("tmp")
     ghost_tree_nwk = open(ghost_tree_fp + "/ghost_tree.nwk", "w")
     ghost_tree_nwk.write(str(foundation_tree))
@@ -246,8 +257,8 @@ def _make_nr_foundation_newick(foundation_fh,
         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_letter + "__" + graft_taxa.lower() +
+                                 ";", foundation_taxonomy.lower()) or
                        re.search(";" + graft_taxa.lower(),
                                  foundation_taxonomy.lower()))
 
@@ -335,7 +346,6 @@ def _collapse(tax, level):
 
 
 def _graft_functions(graft_level):
-    print(graft_level)
     graft_letter = graft_level
     graft_level_map = {'p': 2, 'c': 3, 'o': 4, 'f': 5, 'g': 6}
     graft_level = graft_level_map[graft_letter]

ghosttree/scaffold/tests/test_hybridtree.py

@@ -63,7 +63,8 @@ def test_newick_file_few_extensions(self):
                                             self.graft_letter_g,
                                             self.foundation_taxonomy)
         result = str(result)
-        trimmed_tree = '(CBB3:0.44621,PBB1:0.08710999999999999)0.801:0.14776;\n'
+        trimmed_tree = '(CBB3:0.44621,PBB1:0.08710999999999999)' \
+                       '0.801:0.14776;\n'
         self.assertEqual(result, trimmed_tree)
 
     def test_graft_functions(self):
@@ -82,8 +83,7 @@ def test_create_taxonomy_dic_many_genus(self):
 
     def test_create_taxonomy_dic_none(self):
         with self.assertRaises(ValueError):
-            list(_create_taxonomy_dict(self.extension_taxonomy_none,
-                                       9))
+            list(_create_taxonomy_dict(self.extension_taxonomy_none, 9))
 
     def test_make_mini_otu_files(self):
         os.system("mkdir tmp")

ghosttree/silva/tests/test_filter.py

@@ -35,7 +35,8 @@ def test_fasta_with_fungi(self):
                                   self.accession,
                                   self.taxonomy_with_fungi)
         self.assertEqual(list(result),
-                         [Sequence("ATCG", metadata={'id': "AB21", 'description': "Fungi"})])
+                         [Sequence("ATCG", metadata={'id': "AB21",
+                                                     'description': "Fungi"})])
 
     def test_fasta_with_many_fungi(self):
         result = fungi_from_fasta(self.fasta_many_fungi,

scripts/ghost-tree

@@ -177,22 +177,26 @@ def scaffold():
 @click.argument("ghost-tree-output-folder", type=click.STRING)
 @click.option('--graft-level', type=click.Choice(['p', 'c', 'o', 'f', 'g']),
               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")
+              help="Taxonomic level to graft extensions to foundation. "
+                   "Default is g (genus). p=phylum, c=class, o=order, "
+                   "f=family.")
+@click.option('-i', '--foundation-taxonomy', type=click.File('U'),
+              help="If you use newick tree for your foundation, this is "
+                   "REQUIRED. This is a tab-delimited accession ID and "
+                   "taxonomy line file.")
+@click.option("--stderr", is_flag=True, help="Saves error log file "
+                                             "(optional)")
+@click.option("--save-foundation-alignment", is_flag=True,
+              help="Saves non-redundant foundation alignment file (optional)")
+@click.option("--save-foundation-tree", is_flag=True,
+              help="Saves foundation phylogenetic tree file (optional)")
 @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):
+              help="Will not save accession id file (optional)")
+def hybrid_tree(graft_level, foundation_taxonomy, stderr,
+                save_foundation_alignment, save_foundation_tree,
+                exclude_id_list, extension_trees_otu_map,
+                extension_trees_taxonomy_file, extension_trees_sequence_file,
+                foundation_file, ghost_tree_output_folder):
     """Combines two genetic databases into one phylogenetic tree and is the
     final command in ghost-tree.
 
@@ -205,7 +209,7 @@ def hybrid_tree(graft_level, foundation_taxonomy, stderr, foundation_alignment,
     taxonomic resolution.
 
     The output file is one phylogenetic tree in .nwk format, which has the
-    benefits of both databases, but allows sequencing to be performed using
+    benefits of both databases. This allows sequencing to be performed using
     only one primer set, the "extension trees" primer set.
 
     Arguments:
@@ -218,7 +222,7 @@ def hybrid_tree(graft_level, foundation_taxonomy, stderr, foundation_alignment,
     accession numbers, which are part of that OTU cluster (as in output of
     "ghost-tree group-extensions"). This file refers to the "extension
     trees". File references to sequence reads or sample numbers/names are
-    not valid here. This is not an OTU .biom table.
+    not valid here. This is NOT the same as an OTU .biom table.
 
     2) extension-trees-taxonomy-file:
     Tab-delimited text file related to "extension trees" wih the 1st
@@ -234,20 +238,29 @@ def hybrid_tree(graft_level, foundation_taxonomy, stderr, foundation_alignment,
     accession numbers are the same as in the extension-trees-otu-map and
     extension-trees-taxonomy-file.
 
-    4) foundation-alignment-file:
-    File containing pre-aligned sequences from a genetic marker database
-    in .fasta format. This file refers to the "foundation" of the
-    ghost-tree. Contains accession numbers and taxonomy labels.
+    4) foundation-file:
+    File containing EITHER pre-aligned sequences from a genetic marker
+    database in .fasta format OR a newick tree. This file refers to the
+    "foundation" of the ghost-tree.
+
+    .fasta contains accession numbers *and* taxonomy labels.
+
+    .nwk tree is a tree with accession numbers. MUST supply a foundation
+    taxonomy if using a tree as a foundation. Pass this via the
+    --foundation-taxonomy option. see --help for details
 
     5) ghost-tree-output-folder:
-        Output folder contains files including:
-        a) The Newick formatted ghost-tree, which is the final output of the
-           ghost-tree tool. This is a phylogenetic tree designed for
-           downstream diversity analyses.
-        b) Accession IDs from the ghost-tree.nwk file that you can use for
-           downstream analyses tools
-        c) log error file (this is an optional file that you can have if you
-           type '--stderr')
+    Output folder contains files including:
+
+    a) The Newick formatted ghost-tree, which is the final output of the
+    ghost-tree tool. This is a phylogenetic tree designed for
+    downstream diversity analyses.
+
+    b) Accession IDs from the ghost-tree.nwk file that you can use for
+    downstream analyses tools
+
+    c) log error file (this is an optional file that you can have if you
+    type '--stderr')
     """
     _, std_error = extensions_onto_foundation(extension_trees_otu_map,
                                               extension_trees_taxonomy_file,
@@ -261,7 +274,7 @@ def hybrid_tree(graft_level, foundation_taxonomy, stderr, foundation_alignment,
         logfile.write(std_error)
         logfile.close()
 
-    if foundation_alignment:
+    if save_foundation_alignment:
         pass
     else:
         try:
@@ -270,7 +283,7 @@ def hybrid_tree(graft_level, foundation_taxonomy, stderr, foundation_alignment,
         except FileNotFoundError:
             pass
 
-    if foundation_tree:
+    if save_foundation_tree:
         pass
     else:
         try:

setup.cfg

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