ENH: eggnog_hmmer_search outputs gff files now

q2_moshpit/eggnog/orthologs/common.py

@@ -86,8 +86,8 @@ def _run_eggnog_search_pipeline(
     partition_method = ctx.get_action(plugin, action_name)
     _eggnog_search = ctx.get_action("moshpit", search_action)
     collate_hits = ctx.get_action("types", "collate_orthologs")
-    collate_loci = ctx.get_action("types", "collate_loci")
     _eggnog_feature_table = ctx.get_action("moshpit", "_eggnog_feature_table")
+    collate_loci = ctx.get_action("types", "collate_loci")
     (partitioned_sequences,) = partition_method(sequences, num_partitions)
 
     hits = []

q2_moshpit/eggnog/orthologs/hmmer.py

@@ -19,7 +19,7 @@
 from q2_moshpit.eggnog.types import EggnogHmmerIdmapDirectoryFmt
 from q2_types.feature_data_mag import MAGSequencesDirFmt
 from q2_types.genome_data import (
-    ProteinsDirectoryFormat, SeedOrthologDirFmt
+    ProteinsDirectoryFormat, SeedOrthologDirFmt, LociDirectoryFormat
 )
 from q2_types.per_sample_sequences import (
     ContigSequencesDirFmt, MultiMAGSequencesDirFmt
@@ -38,7 +38,7 @@ def _eggnog_hmmer_search(
     seed_alignments: ProteinsDirectoryFormat,
     num_cpus: int = 1,
     db_in_memory: bool = False
-) -> (SeedOrthologDirFmt, pd.DataFrame):
+) -> (SeedOrthologDirFmt, pd.DataFrame, LociDirectoryFormat):
     with tempfile.TemporaryDirectory() as output_loc:
         taxon_id = os.listdir(idmap.path)[0].split(".")[0]
         tmp_subdir = f"{output_loc}/hmmer/{taxon_id}"
@@ -54,17 +54,18 @@ def _eggnog_hmmer_search(
                 '--genepred', 'prodigal'  # default incompatible with HMMER
             ]
         )
-        result, ft = _eggnog_search(sequences, search_runner, output_loc)
-    return result, ft
+        result, ft, loci = _eggnog_search(sequences, search_runner, output_loc)
+    return result, ft, loci
 
 
 def eggnog_hmmer_search(
-    ctx, sequences, pressed_hmm_db, idmap, seed_alignments,
-    num_cpus=1, db_in_memory=False, num_partitions=None
+        ctx, sequences, pressed_hmm_db, idmap, seed_alignments,
+        num_cpus=1, db_in_memory=False, num_partitions=None
 ):
-    collated_hits, collated_tables = _run_eggnog_search_pipeline(
-        ctx, sequences, [idmap, pressed_hmm_db, seed_alignments],
-        num_cpus, db_in_memory, num_partitions,
-        "_eggnog_hmmer_search"
-    )
-    return collated_hits, collated_tables
+    collated_hits, collated_tables, collated_loci = (
+        _run_eggnog_search_pipeline(
+            ctx, sequences, [idmap, pressed_hmm_db, seed_alignments],
+            num_cpus, db_in_memory, num_partitions,
+            "_eggnog_hmmer_search"
+        ))
+    return collated_hits, collated_tables, collated_loci

q2_moshpit/eggnog/tests/test_orthologs.py

@@ -73,9 +73,11 @@ def setUp(self):
     def test_eggnog_hmmer_search_pipeline(self):
         mock_action = MagicMock(side_effect=[
             lambda sequences, num_partitions: ({"mag1": {}, "mag2": {}},),
-            lambda seq, pressed, idmap, fastas, num_cpus, db_in_memory: (0, 0),
+            lambda seq, pressed, idmap, fastas, num_cpus, db_in_memory:
+            (0, 0, 0),
             lambda hits: ("collated_hits",),
             lambda collated_hits: ("collated_tables",),
+            lambda collated_loci: ("collated_loci",),
         ])
         mock_ctx = MagicMock(get_action=mock_action)
         obs = eggnog_hmmer_search(
@@ -85,7 +87,7 @@ def test_eggnog_hmmer_search_pipeline(self):
             idmap=self.idmap_artifact,
             seed_alignments=self.fastas_artifact
         )
-        exp = ("collated_hits", "collated_tables")
+        exp = ("collated_hits", "collated_tables", "collated_loci")
         self.assertTupleEqual(obs, exp)
 
     def test_symlink_files_to_target_dir(self):
@@ -114,8 +116,8 @@ def test_eggnog_hmmer_search(
             self, mock_eggnog_search, mock_symlink, mock_tmpdir, mock_makedirs
     ):
         mock_tmpdir.return_value.__enter__.return_value = "tmp"
-        mock_eggnog_search.return_value = (0, 1)
-        result, ft = _eggnog_hmmer_search(
+        mock_eggnog_search.return_value = (0, 1, 2)
+        result, ft, loci = _eggnog_hmmer_search(
             sequences=self.mags,
             idmap=self.idmap,
             pressed_hmm_db=self.pressed_hmm,
@@ -129,7 +131,7 @@ def test_eggnog_hmmer_search(
             ANY,  # partial() method not patchable or comparable
             "tmp"
         )
-        self.assertTupleEqual((result, ft), (0, 1))
+        self.assertTupleEqual((result, ft, loci), (0, 1, 2))
 
     def test_eggnog_search_mags(self):
         sequences = MultiMAGSequencesDirFmt(
@@ -138,7 +140,7 @@ def test_eggnog_search_mags(self):
         output_loc = self.get_data_path('hits')
         search_runner = MagicMock()
 
-        result, ft = _eggnog_search(sequences, search_runner, output_loc)
+        result, ft, _ = _eggnog_search(sequences, search_runner, output_loc)
         result.validate()
         self.assertIsInstance(ft, pd.DataFrame)
 
@@ -155,7 +157,7 @@ def test_eggnog_search_contigs(self):
         output_loc = self.get_data_path('hits')
         search_runner = MagicMock()
 
-        result, ft = _eggnog_search(sequences, search_runner, output_loc)
+        result, ft, _ = _eggnog_search(sequences, search_runner, output_loc)
         result.validate()
         self.assertIsInstance(ft, pd.DataFrame)
 
@@ -171,7 +173,7 @@ def test_eggnog_search_mags_derep(self):
         output_loc = self.get_data_path('hits')
         search_runner = MagicMock()
 
-        result, ft = _eggnog_search(sequences, search_runner, output_loc)
+        result, ft, _ = _eggnog_search(sequences, search_runner, output_loc)
         result.validate()
         self.assertIsInstance(ft, pd.DataFrame)
 

q2_moshpit/plugin_setup.py

@@ -736,7 +736,8 @@
     },
     outputs=[
         ('eggnog_hits', SampleData[Orthologs]),
-        ('table', FeatureTable[Frequency])
+        ('table', FeatureTable[Frequency]),
+        ('loci', GenomeData[Loci]),
     ],
     name='Run eggNOG search using HMMER aligner',
     description="This method uses HMMER to find possible target sequences "
@@ -824,12 +825,14 @@
     },
     outputs=[
         ('eggnog_hits', SampleData[Orthologs]),
-        ('table', FeatureTable[Frequency])
+        ('table', FeatureTable[Frequency]),
+        ('loci', GenomeData[Loci])
     ],
     output_descriptions={
         'eggnog_hits': 'BLAST6-like table(s) describing the identified '
                        'orthologs. One table per sample or MAG in the input.',
-        'table': 'Feature table with counts of orthologs per sample/MAG.'
+        'table': 'Feature table with counts of orthologs per sample/MAG.',
+        'loci': 'Loci of the identified orthologs.'
     },
     name='Run eggNOG search using HMMER aligner',
     description='This method performs the steps by which we find our '