smaller updates

McTavishLab · Nov 27, 2018 · 1093d92 · 1093d92
1 parent cf0f15d
commit 1093d92
Show file tree

Hide file tree

Showing 4 changed files with 10 additions and 6 deletions.
diff --git a/physcraper/__init__.py b/physcraper/__init__.py
@@ -2177,11 +2177,11 @@ def remove_alien_aln_tre(self):
         This function ensures they are properly removed."""
         for tax_lab in self.data.aln.taxon_namespace:
             if tax_lab not in self.data.tre.taxon_namespace:
-                sys.stdout.write("tax not in tre. This is an alien name in the data.")
+                sys.stderr.write("tax {} not in tre. This is an alien name in the data.\n".format(tax_lab))
                 self.data.remove_taxa_aln_tre(tax_lab)
         for tax_lab in self.data.tre.taxon_namespace:
             if tax_lab not in self.data.aln.taxon_namespace:
-                sys.stdout.write("tax not in aln. This is an alien name in the data. ")
+                sys.stderr.write("tax {} not in aln. This is an alien name in the data.\n".format(tax_lab))
                 self.data.remove_taxa_aln_tre(tax_lab)
         self.data.prune_short()
 
@@ -2368,6 +2368,7 @@ def generate_streamed_alignment(self):
                 self.write_query_seqs()
                 self.align_query_seqs()
                 self.place_query_seqs()
+                self.prune_short()
                 self.est_full_tree()
                 self.data.tre = Tree.get(path="{}/RAxML_bestTree.{}".format(self.workdir, self.date),
                                          schema="newick",
@@ -2532,6 +2533,8 @@ def sp_dict(self, downtorank=None):
                     if tax_name is None:
                         debug("something is going wrong!Check species name")
                         sys.stderr.write("{} has no corresponding tax_name! Check what is wrong!".format(key))
+                if len(tax_name.split("(")) > 1:
+                    tax_name = tax_name.split("(")[0]
                 tax_name = str(tax_name).replace(" ", "_")
                 if self.config.blast_loc == 'remote':
                     if '^ncbi:accession' in self.data.otu_dict[key]:
@@ -2958,11 +2961,12 @@ def write_otu_info(self, downtorank=None):
             "^ot:ottTaxonName",
             "^ncbi:gi",
             "^ncbi:accession",
+            "^ot:originalLabel",
             "^physcraper:last_blasted",
             "^physcraper:status",
             "^ot:ottId",
             "^ncbi:taxon",
-            "^ncbi:title",
+            "^ncbi:title"
         ]
         with open("{}/otu_seq_info.csv".format(self.workdir), "w") as output:
             writer = csv.writer(output)

diff --git a/tests/tiny_comb_ets.py b/tests/tiny_comb_ets.py
@@ -11,7 +11,7 @@
 id_to_spn = r"tests/data/tiny_comb_ets/nicespl.csv"
 
 workdir = "tiny_comb_ets"
-configfi = "tests/data/test.config"
+configfi = "tests/data/localblast.config"
 otu_jsonfi = "{}/otu_dict.json".format(workdir)
 threshold = 2
 selectby = "blast"

diff --git a/tests/tiny_comb_its.py b/tests/tiny_comb_its.py
@@ -11,7 +11,7 @@
 id_to_spn = r"tests/data/tiny_comb_its/nicespl.csv"
 
 workdir = "tiny_comb_its"
-configfi = "tests/data/test.config"
+configfi = "tests/data/localblast.config"
 otu_jsonfi = "{}/otu_dict.json".format(workdir)
 threshold = 2
 selectby = "blast"

diff --git a/tests/tiny_filter_ownfile.py b/tests/tiny_filter_ownfile.py
@@ -10,7 +10,7 @@
 schema_trf = "newick"
 id_to_spn = r"tests/data/tiny_test_example/test_nicespl.csv"
 workdir="tests/output/tiny_filter_own"
-configfi = "tests/data/test.config"
+configfi = "tests/data/localblast.config"
 
 otu_jsonfi = "{}/otu_dict.json".format(workdir)