Add hmmer-domtab indexing and its tests

Bio/SearchIO/HmmerIO/hmmerdomtab.py

@@ -10,6 +10,7 @@
 from Bio._py3k import _as_bytes, _bytes_to_string
 from Bio.SearchIO._objects import QueryResult, Hit, HSP
 from Bio.SearchIO._index import SearchIndexer
+from hmmertab import HmmerTabIndexer
 
 
 def hmmer_domtab_hmmhit_iterator(handle):
@@ -183,48 +184,28 @@ def parse_qresult(self):
             self.line = read_forward(self.handle)
 
 
-class HmmerDomtabHmmhitIndexer(SearchIndexer):
+class HmmerDomtabHmmhitIndexer(HmmerTabIndexer):
 
     """Indexer class for HMMER domain table output that assumes HMM profile
     coordinates are hit coordinates."""
 
     def __init__(self, *args, **kwargs):
-        SearchIndexer.__init__(self, *args, **kwargs)
+        HmmerTabIndexer.__init__(self, *args, **kwargs)
         # set parser for on-the-fly parsing
-        self._parser = hmmer_domtab_iterator
-        self._handle.seek(0)
+        self._parser = hmmer_domtab_hmmhit_iterator
+        self._query_id_idx = 3
 
-    def __iter__(self):
-        """Iterates over the file handle; yields key, start offset, and length."""
-        handle = self._handle
-        handle.seek(0)
-        split_char = _as_bytes(' ')
-        qresult_key = None
-
-        # read through header
-        while True:
-            start_offset = handle.tell()
-            line = read_forward(handle, strip=False)
-            if not line.startswith('#'):
-                break
 
-        # and index the qresults
-        #while True:
-        #    end_offset = handle.tell()
+class HmmerDomtabHmmqueryIndexer(HmmerTabIndexer):
 
-        #    if not line:
-        #        break
-        #    if qresult_key is None:
-
-    def get_raw(self, offset):
-        """Returns the raw string of a QueryResult object from the given offset."""
-        handle = self._handle
-        handle.seek(offset)
-        split_char = _as_bytes(' ')
-        qresult_key = None
-        qresult_raw = ''
+    """Indexer class for HMMER domain table output that assumes HMM profile
+    coordinates are query coordinates."""
 
-        return qresult_raw
+    def __init__(self, *args, **kwargs):
+        HmmerTabIndexer.__init__(self, *args, **kwargs)
+        # set parser for on-the-fly parsing
+        self._parser = hmmer_domtab_hmmquery_iterator
+        self._query_id_idx = 3
 
 
 class HmmerDomtabHmmhitWriter(object):

Bio/SearchIO/HmmerIO/hmmertab.py

@@ -152,12 +152,15 @@ def __init__(self, *args, **kwargs):
         # set parser for on-the-fly parsing
         self._parser = hmmer_tab_iterator
         self._handle.seek(0)
+        # denotes column location for query identifier
+        self._query_id_idx = 2
 
     def __iter__(self):
         """Iterates over the file handle; yields key, start offset, and length."""
         handle = self._handle
         handle.seek(0)
         split_char = _as_bytes(' ')
+        query_id_idx = self._query_id_idx
         qresult_key = None
 
         # read through header
@@ -174,9 +177,11 @@ def __iter__(self):
             if not line:
                 break
             if qresult_key is None:
-                qresult_key = filter(None, line.strip().split(split_char))[2]
+                qresult_key = filter(None, \
+                        line.strip().split(split_char))[query_id_idx]
             else:
-                curr_key = filter(None, line.strip().split(split_char))[2]
+                curr_key = filter(None, \
+                        line.strip().split(split_char))[query_id_idx]
 
                 if curr_key != qresult_key:
                     yield _bytes_to_string(qresult_key), start_offset, \
@@ -195,6 +200,7 @@ def get_raw(self, offset):
         handle = self._handle
         handle.seek(offset)
         split_char = _as_bytes(' ')
+        query_id_idx = self._query_id_idx
         qresult_key = None
         qresult_raw = ''
 
@@ -203,9 +209,11 @@ def get_raw(self, offset):
             if not line:
                 break
             if qresult_key is None:
-                qresult_key = filter(None, line.strip().split(split_char))[2]
+                qresult_key = filter(None, \
+                        line.strip().split(split_char))[query_id_idx]
             else:
-                curr_key = filter(None, line.strip().split(split_char))[2]
+                curr_key = filter(None, \
+                        line.strip().split(split_char))[query_id_idx]
                 if curr_key != qresult_key:
                     break
             qresult_raw += line

Tests/test_SearchIO_index.py

@@ -1189,6 +1189,50 @@ def test_hmmertab_30_single(self):
         self.assertEqual(raw, idx.get_raw('gi|126362951:116-221'))
 
 
+class HmmerDomtabRawCases(unittest.TestCase):
+
+    fmt = 'hmmscan-domtab'
+
+    def test_hmmerdomtab_30_multiple_first(self):
+        """Test hmmscan-domtab raw string retrieval, HMMER 3.0, multiple queries, first (domtab_30_hmmscan_001.out)"""
+        filename = os.path.join('Hmmer', 'domtab_30_hmmscan_001.out')
+        idx = SearchIO.index(filename, self.fmt)
+        raw = """Globin               PF00042.17   108 gi|4885477|ref|NP_005359.1| -            154     6e-21   74.6   0.3   1   1   6.7e-25   9.2e-21   74.0   0.2     1   107     7   112     7   113 0.97 Globin
+"""
+        self.assertEqual(raw, idx.get_raw('gi|4885477|ref|NP_005359.1|'))
+
+    def test_hmmerdomtab_30_multiple_middle(self):
+        """Test hmmscan-domtab raw string retrieval, HMMER 3.0, multiple queries, middle (domtab_30_hmmscan_001.out)"""
+        filename = os.path.join('Hmmer', 'domtab_30_hmmscan_001.out')
+        idx = SearchIO.index(filename, self.fmt)
+        raw = """Ig_3                 PF13927.1     75 gi|126362951:116-221 -            106   1.4e-09   38.2   0.4   1   1     3e-13   2.1e-09   37.6   0.3     1    73     9    84     9    88 0.94 Immunoglobulin domain
+Ig_2                 PF13895.1     80 gi|126362951:116-221 -            106   3.5e-05   23.7   0.1   1   1   6.2e-09   4.3e-05   23.4   0.1     1    80     9   104     9   104 0.71 Immunoglobulin domain
+"""
+        self.assertEqual(raw, idx.get_raw('gi|126362951:116-221'))
+
+    def test_hmmerdomtab_30_multiple_last(self):
+        """Test hmmscan-domtab raw string retrieval, HMMER 3.0, multiple queries, last (domtab_30_hmmscan_001.out)"""
+        filename = os.path.join('Hmmer', 'domtab_30_hmmscan_001.out')
+        idx = SearchIO.index(filename, self.fmt)
+        raw = """Pou                  PF00157.12    75 gi|125490392|ref|NP_038661.2| -            352     7e-37  124.8   0.5   1   1     5e-40   1.4e-36  123.9   0.3     3    75   133   205   131   205 0.97 Pou domain - N-terminal to homeobox domain
+Homeobox             PF00046.24    57 gi|125490392|ref|NP_038661.2| -            352   2.1e-18   65.5   1.1   1   1   1.5e-21   4.1e-18   64.6   0.7     1    57   224   280   224   280 0.98 Homeobox domain
+HTH_31               PF13560.1     64 gi|125490392|ref|NP_038661.2| -            352     0.012   15.6   0.0   1   2   5.7e-05      0.16   12.0   0.0     1    35   141   181   141   184 0.96 Helix-turn-helix domain
+HTH_31               PF13560.1     64 gi|125490392|ref|NP_038661.2| -            352     0.012   15.6   0.0   2   2      0.19   5.2e+02    0.8   0.0    39    62   245   268   243   270 0.86 Helix-turn-helix domain
+Homeobox_KN          PF05920.6     40 gi|125490392|ref|NP_038661.2| -            352     0.039   13.5   0.0   1   1   3.5e-05     0.095   12.3   0.0     7    39   244   276   241   277 0.91 Homeobox KN domain
+DUF521               PF04412.8    400 gi|125490392|ref|NP_038661.2| -            352      0.14   10.5   0.1   1   1   9.4e-05      0.26    9.6   0.1   273   334   221   280   197   294 0.77 Protein of unknown function (DUF521)
+"""
+        self.assertEqual(raw, idx.get_raw('gi|125490392|ref|NP_038661.2|'))
+
+    def test_hmmerdomtab_30_single(self):
+        """Test hmmscan-domtab raw string retrieval, HMMER 3.0, single query (domtab_30_hmmscan_004.out)"""
+        filename = os.path.join('Hmmer', 'domtab_30_hmmscan_004.out')
+        idx = SearchIO.index(filename, self.fmt)
+        raw = """Ig_3                 PF13927.1     75 gi|126362951:116-221 -            106   1.4e-09   38.2   0.4   1   1     3e-13   2.1e-09   37.6   0.3     1    73     9    84     9    88 0.94 Immunoglobulin domain
+Ig_2                 PF13895.1     80 gi|126362951:116-221 -            106   3.5e-05   23.7   0.1   1   1   6.2e-09   4.3e-05   23.4   0.1     1    80     9   104     9   104 0.71 Immunoglobulin domain
+"""
+        self.assertEqual(raw, idx.get_raw('gi|126362951:116-221'))
+
+
 class SearchIndexCases(unittest.TestCase):
 
     def check_index(self, filename, format):
@@ -1359,6 +1403,34 @@ def test_hmmertab_30_hmmscan_004(self):
         self.check_index(filename, self.fmt)
 
 
+class HmmerDomtabIndexCases(SearchIndexCases):
+
+    def test_hmmerdomtab_30_hmmscan_001(self):
+        """Test hmmscan-domtab indexing, HMMER 3.0, multiple queries"""
+        filename = os.path.join('Hmmer', 'domtab_30_hmmscan_001.out')
+        self.check_index(filename, 'hmmscan-domtab')
+
+    def test_hmmerdomtab_30_hmmscan_002(self):
+        """Test hmmscan-domtab indexing, HMMER 3.0, single query, no hits"""
+        filename = os.path.join('Hmmer', 'domtab_30_hmmscan_002.out')
+        self.check_index(filename, 'hmmscan-domtab')
+
+    def test_hmmerdomtab_30_hmmscan_003(self):
+        """Test hmmscan-domtab indexing, HMMER 3.0, single query, multiple hits"""
+        filename = os.path.join('Hmmer', 'domtab_30_hmmscan_003.out')
+        self.check_index(filename, 'hmmscan-domtab')
+
+    def test_hmmerdomtab_30_hmmscan_004(self):
+        """Test hmmscan-domtab indexing, HMMER 3.0, single query, no alignments"""
+        filename = os.path.join('Hmmer', 'domtab_30_hmmscan_004.out')
+        self.check_index(filename, 'hmmscan-domtab')
+
+    def test_hmmerdomtab_30_hmmsearch_001(self):
+        """Test hmmsearch-domtab indexing, HMMER 3.0, single query, no alignments"""
+        filename = os.path.join('Hmmer', 'domtab_30_hmmsearch_001.out')
+        self.check_index(filename, 'hmmsearch-domtab')
+
+
 if __name__ == "__main__":
     runner = unittest.TextTestRunner(verbosity = 2)
     unittest.main(testRunner=runner)