Merge pull request HadrienG#25 from HadrienG/lineage

Lineage

.travis.yml

@@ -1,8 +1,11 @@
 language: python
+dist: xenial
+sudo: true
 python:
 - '3.5'
+- '3.6'
+- '3.7'
 install:
-- pip install -r requirements.txt
 - pip install nose nose-testconfig
 - pip install codecov
 - pip install -e .

Pipfile

@@ -0,0 +1,17 @@
+[[source]]
+url = "https://pypi.python.org/simple"
+verify_ssl = true
+name = "pypi"
+
+[packages]
+requests = "*"
+peewee = "==2.8.1"
+tqdm = "*"
+
+[dev-packages]
+nose = "*"
+nose-testconfig = "*"
+
+[scripts]
+tests = "nosetests --tc-file taxadb.ini --with-coverage --cover-package=taxadb"
+taxadb = "python -m taxadb"

taxadb/app.py

@@ -293,9 +293,9 @@ def main():
     parser_create.add_argument(
         '--division',
         '-d',
-        choices=['full', 'nucl', 'prot', 'gb', 'wgs', 'gss', 'est'],
+        choices=['taxa', 'full', 'nucl', 'prot', 'gb', 'wgs', 'gss', 'est'],
         default='full',
-        metavar='[full|nucl|prot|gb|wgs|gss|est]',
+        metavar='[taxa|full|nucl|prot|gb|wgs|gss|est]',
         help='division to build (default: %(default)s))'
     )
     parser_create.add_argument(
@@ -355,4 +355,4 @@ def main():
         logger = logging.getLogger(__name__)
         logger.debug(e)
         parser.print_help()
-        raise  # extra traceback to uncomment for extra debugging powers
+        # raise  # extra traceback to uncomment for extra debugging powers

taxadb/taxid.py

@@ -38,14 +38,16 @@ def sci_name(self, taxid):
         except Taxa.DoesNotExist:
             return None
 
-    def lineage_id(self, taxid, reverse=False):
+    def lineage_id(self, taxid, ranks=False, reverse=False):
         """Get lineage for a taxonomic id
 
         Given a taxid, return its associated lineage (in the form of a list of
             taxids, each parents of each others)
 
         Args:
             taxid (:obj:`int`): a taxid
+            ranks (:obj:`bool`): Wether to return a dict with the tax ranks or
+                not. Default False
             reverse (:obj:`bool`): Inverted lineage, from top to bottom
                 taxonomy hierarchy. Default False
         Returns:
@@ -54,30 +56,39 @@ def lineage_id(self, taxid, reverse=False):
 
         """
         try:
-            lineage_list = []
+            lineages = {} if ranks else []
+            # lineage_list = []
             current_lineage = Taxa.get(Taxa.ncbi_taxid == taxid).tax_name
             current_lineage_id = Taxa.get(Taxa.ncbi_taxid == taxid).ncbi_taxid
             parent = Taxa.get(Taxa.ncbi_taxid == taxid).parent_taxid
+            rank = Taxa.get(Taxa.ncbi_taxid == taxid).lineage_level
             while current_lineage != 'root':
-                lineage_list.append(current_lineage_id)
+                if ranks:
+                    lineages[rank] = current_lineage_id
+                else:
+                    lineages.append(current_lineage_id)
+
                 new_query = Taxa.get(Taxa.ncbi_taxid == parent)
 
                 current_lineage = new_query.tax_name
                 current_lineage_id = new_query.ncbi_taxid
                 parent = new_query.parent_taxid
-            if reverse is True:
-                lineage_list.reverse()
-            return lineage_list
+                rank = new_query.lineage_level
+            if reverse is True and ranks is False:
+                lineages.reverse()
+            return lineages
         except Taxa.DoesNotExist:
             return None
 
-    def lineage_name(self, taxid, reverse=False):
+    def lineage_name(self, taxid, ranks=False, reverse=False):
         """Get a lineage name for a taxonomic id
 
         Given a taxid, return its associated lineage
 
         Arguments:
             taxid (:obj:`int`): a taxid
+            ranks (:obj:`bool`): Wether to return a dict with the tax ranks or
+                not. Default False
             reverse (:obj:`bool`): Inverted lineage, from top to bottom
                 taxonomy hierarchy. Default False
 
@@ -87,18 +98,23 @@ def lineage_name(self, taxid, reverse=False):
 
         """
         try:
-            lineage_list = []
+            lineages = {} if ranks else []
             current_lineage = Taxa.get(Taxa.ncbi_taxid == taxid).tax_name
             parent = Taxa.get(Taxa.ncbi_taxid == taxid).parent_taxid
+            rank = Taxa.get(Taxa.ncbi_taxid == taxid).lineage_level
             while current_lineage != 'root':
-                lineage_list.append(current_lineage)
+                if ranks:
+                    lineages[rank] = current_lineage
+                else:
+                    lineages.append(current_lineage)
                 new_query = Taxa.get(Taxa.ncbi_taxid == parent)
 
                 current_lineage = new_query.tax_name
                 parent = new_query.parent_taxid
-            if reverse is True:
-                lineage_list.reverse()
-            return lineage_list
+                rank = new_query.lineage_level
+            if reverse is True and ranks is False:
+                lineages.reverse()
+            return lineages
         except Taxa.DoesNotExist:
             return None
 

taxadb/test/test_taxadb.py

@@ -388,13 +388,15 @@ def test_taxid_sci_name_None(self):
         self.assertIsNone(name)
 
     @attr('taxid')
-    def test_taxid_lineage_id(self):
+    def test_taxid_lineage_id_ranks(self):
         taxid = self._buildTaxaDBObject(TaxID)
-        lineage = taxid.lineage_id(9986)
-        self.assertListEqual(lineage, [
-            9986, 9984, 9979, 9975, 314147, 314146, 1437010, 9347, 32525,
-            40674, 32524, 32523, 1338369, 8287, 117571, 117570, 7776, 7742,
-            89593, 7711, 33511, 33213, 6072, 33208, 33154, 2759, 131567])
+        lineage = taxid.lineage_id(9986, ranks=True)
+        self.assertDictEqual(lineage,
+                             {'species': 9986, 'genus': 9984, 'family': 9979,
+                              'order': 9975, 'no rank': 131567,
+                              'superorder': 314146, 'class': 40674,
+                              'subphylum': 89593, 'phylum': 7711,
+                              'kingdom': 33208, 'superkingdom': 2759})
 
     @attr('taxid')
     def test_taxid_lineage_id_reverse(self):

taxadb/version.py

@@ -1 +1 @@
-__version__ = '0.8.0'
+__version__ = '0.9.0'