Add available name format and improve error report

phylopypruner/msa.py

@@ -59,17 +59,16 @@ def add_sequence(self, seq_record=None, description="", sequence_data=""):
             seq_record = sequence.Sequence()
             seq_record.description = description
             seq_record.sequence_data = sequence_data
-            seq_record.otu = re.split(r"\||@", seq_record.description)[0]
+            seq_record.otu = re.split(r"\||@|_", seq_record.description)[0]
             try:
-                seq_record.identifier = re.split(
-                    r"\||@", seq_record.description)[1]
-            except IndexError:
-                report.warning("no description found on split with | or @")
+                seq_record.identifier = re.search(r"[|@_]([^ ]*)", seq_record.description).group(1)
+            except AttributeError:
+                report.warning("no description found on split with | , _ or @")
                 seq_record.identifier = None
         if description:
-            seq_record.otu = re.split(r"\||@", seq_record.description)[0]
+            seq_record.otu = re.split(r"\||@|_", seq_record.description)[0]
         if sequence_data:
-            seq_record.identifier = re.split(r"\||@", seq_record.description)[1]
+            seq_record.identifier = re.search(r"[|@_]([^ ]*)", seq_record.description).group(1)
 
         self.sequences.append(seq_record)
         return seq_record

phylopypruner/run.py

@@ -16,14 +16,17 @@
 
 def validate_input(msa, tree, tree_path):
     "Test to see if MSA and tree entries matches."
-    descriptions = list(msa.iter_descriptions())
-    names = list(tree.iter_names())
-
-    if set(descriptions).intersection(names) < set(descriptions):
-        print("example tree names:", names[:2], file=sys.stderr)
-        print("example sequences:", descriptions[:2], file=sys.stderr)
-        report.error("MSA names don't match tree \n   {}\n   {}".format(
-            msa.filename, tree_path))
+    descriptions = set(msa.iter_descriptions())
+    names = set(tree.iter_names())
+
+    # Find items in descriptions not in names and vice versa
+    missing_in_names = descriptions - names
+    missing_in_descriptions = names - descriptions
+
+    if missing_in_names or missing_in_descriptions:
+        print("Missing in tree names:", list(missing_in_names), file=sys.stderr)
+        print("Missing in sequences:", list(missing_in_descriptions), file=sys.stderr)
+        report.error("MSA names don't match tree \n   {}\n   {}".format(msa.filename, tree_path))
 
 
 def run(settings, msa, tree):

phylopypruner/sequence.py

@@ -21,7 +21,8 @@ def __init__(self, description="", sequence_data=""):
         self._sequence_data = str(sequence_data)
         self._is_alignment = bool(self.is_alignment)
         if description:
-            self._otu, self._identifier = re.split(r"\||@", description)
+            self._otu = re.split(r"\||@|_", description)[0]
+            self._identifier = re.search(r"[|@_]([^ ]*)", description).group(1)
         else:
             self._otu = ""
             self._identifier = ""

phylopypruner/tree_node.py

@@ -108,7 +108,7 @@ def iter_sisters(self):
 
     def otu(self):
         "Returns the OTU to which this node belongs."
-        return re.split(r"\||@", self.name)[0]
+        return re.split(r"\||@|_", self.name)[0]
 
     def is_root(self):
         "Returns True if this node lacks a parent."
@@ -526,15 +526,15 @@ def iter_names(self):
     def iter_otus(self):
         "Returns an iterator object that includes all OTUs within this node."
         for name in self.iter_names():
-            otu = re.split(r"\||@", name)[0]
+            otu = re.split(r"\||@|_", name)[0]
             yield otu
 
     def iter_identifiers(self):
         """
         Returns an iterator object that includes all identifiers in this node.
         """
         for name in self.iter_names():
-            identifier = re.split(r"\||@", name)[1]
+            identifier = re.search(r"[|@_]([^ ]*)", name).group(1)
             yield identifier
 
     def view(self):