feat: add tax_id parameter

htsinfer/exceptions.py

@@ -44,3 +44,9 @@ class TranscriptsFastaProblem(Exception):
 
 class CutadaptProblem(Exception):
     """Exception raised when running cutadapt commands."""
+
+
+class UnsupportedSampleSourceException(Exception):
+    """Exception raised when taxonomy ID is not found in the source
+    organism list.
+    """

htsinfer/get_library_source.py

@@ -4,7 +4,6 @@
 from pathlib import Path
 import subprocess as sp
 import tempfile
-from typing import Optional
 
 from Bio import SeqIO  # type: ignore
 import pandas as pd  # type: ignore
@@ -14,6 +13,7 @@
     FileProblem,
     KallistoProblem,
     TranscriptsFastaProblem,
+    UnsupportedSampleSourceException,
 )
 from htsinfer.models import (
     ResultsSource,
@@ -82,34 +82,11 @@ def evaluate(self) -> ResultsSource:
                 self.tax_id,
                 self.transcripts_file
             )
+            source.file_1.short_name = src_name
 
-            if src_name is not None:
-                source.file_1.short_name = src_name
-
-                if self.paths[1] is not None:
-                    source.file_2.taxon_id = self.tax_id
-                    source.file_2.short_name = source.file_1.short_name
-
-            else:
-                LOGGER.warning(
-                    f"Taxon ID '{self.tax_id}' not found in "
-                    "organism dictionary, inferring source organism..."
-                )
-                index = self.create_kallisto_index()
-                library_source = self.get_source(
-                    fastq=self.paths[0],
-                    index=index,
-                )
-                source.file_1.short_name = library_source.short_name
-                source.file_1.taxon_id = library_source.taxon_id
-
-                if self.paths[1] is not None:
-                    library_source = self.get_source(
-                        fastq=self.paths[1],
-                        index=index,
-                    )
-                    source.file_2.short_name = library_source.short_name
-                    source.file_2.taxon_id = library_source.taxon_id
+            if self.paths[1] is not None:
+                source.file_2.taxon_id = self.tax_id
+                source.file_2.short_name = source.file_1.short_name
 
         else:
             index = self.create_kallisto_index()
@@ -333,7 +310,7 @@ def get_source_expression(
     def get_source_name(
         taxon_id: int,
         transcripts_file: Path,
-    ) -> Optional[str]:
+    ) -> str:
         """Return name of the source organism, based on tax ID.
 
         Args:
@@ -344,10 +321,11 @@ def get_source_name(
             Short name of the organism belonging to the given tax ID.
 
         Raises:
-            Could not process input FASTA file.
+            FileProblem: Could not process input FASTA file.
+            UnsupportedSampleSourceException: Taxon ID is not supported.
         """
         src_dict = {}
-        # Construct dictionary of taxonomy ID's and short names
+
         try:
             for record in list(SeqIO.parse(
                     handle=transcripts_file,
@@ -363,7 +341,10 @@ def get_source_name(
                 f"Could not process file '{transcripts_file}'"
             ) from exc
 
-        if taxon_id in src_dict:
+        try:
             return src_dict[taxon_id]
 
-        return None
+        except KeyError as exc:
+            raise UnsupportedSampleSourceException(
+                f'Taxon ID "{taxon_id}" is not supported by HTSinfer.'
+            ) from exc

tests/test_get_library_source.py

@@ -5,7 +5,8 @@
 from htsinfer.exceptions import (
     FileProblem,
     KallistoProblem,
-    TranscriptsFastaProblem
+    TranscriptsFastaProblem,
+    UnsupportedSampleSourceException,
 )
 from htsinfer.get_library_source import GetLibSource
 from htsinfer.models import (
@@ -323,36 +324,36 @@ def test_create_kallisto_index_problem(self, tmpdir):
         with pytest.raises(KallistoProblem):
             test_instance.create_kallisto_index()
 
-    def test_get_organism_name_found(self):
+    def test_get_source_name_found(self):
         """Test the function when the taxon_id
         is found in the organism dictionary."""
         CONFIG.args.t_file_processed = FILE_TRANSCRIPTS
         test_instance = GetLibSource(config=CONFIG)
         taxon_id = 7227
-        result = test_instance.get_organism_name(
+        result = test_instance.get_source_name(
             taxon_id, CONFIG.args.t_file_processed
         )
         assert result == "dmelanogaster"
 
-    def test_get_organism_name_not_found(self):
+    def test_get_source_name_not_found(self):
         """Test the function when the taxon_id
         is not found in the organism dictionary."""
         CONFIG.args.t_file_processed = FILE_TRANSCRIPTS
         test_instance = GetLibSource(config=CONFIG)
         taxon_id = 12345  # A tax ID that doesn't exist in transcripts
-        result = test_instance.get_organism_name(
-            taxon_id, CONFIG.args.t_file_processed
-        )
-        assert result is None
+        with pytest.raises(UnsupportedSampleSourceException):
+            test_instance.get_source_name(
+                taxon_id, CONFIG.args.t_file_processed
+            )
 
-    def test_get_organism_name_file_problem(self):
+    def test_get_source_name_file_problem(self):
         """Test the function when there's a
         file problem while processing the FASTA file."""
         CONFIG.args.t_file_processed = FILE_DUMMY
         test_instance = GetLibSource(config=CONFIG)
         taxon_id = 7227
         with pytest.raises(FileProblem):
-            test_instance.get_organism_name(
+            test_instance.get_source_name(
                 taxon_id, CONFIG.args.t_file_processed
             )
 
@@ -386,42 +387,6 @@ def test_evaluate_tax_id_is_none(self, monkeypatch, tmpdir):
         assert result.file_2.taxon_id == SOURCE_FRUIT_FLY.taxon_id
         assert result.file_2.short_name == SOURCE_FRUIT_FLY.short_name
 
-    def test_evaluate_tax_id_not_none_no_src_name(self, monkeypatch, tmpdir):
-        """Test when self.tax_id is not None but src_name is not found."""
-        CONFIG.args.tax_id = 7227
-        CONFIG.args.path_1_processed = FILE_MATE_1
-        CONFIG.args.path_2_processed = FILE_MATE_2
-        CONFIG.args.t_file_processed = FILE_TRANSCRIPTS
-        CONFIG.args.tmp_dir = tmpdir
-        CONFIG.args.out_dir = tmpdir
-        test_instance = GetLibSource(config=CONFIG)
-
-        # Mock the get_organism_name method to return None
-        monkeypatch.setattr(
-            'htsinfer.get_library_source.GetLibSource.get_organism_name',
-            lambda *args, **kwargs: None,
-        )
-
-        # Mock the create_kallisto_index method to return a specific result
-        monkeypatch.setattr(
-            'htsinfer.get_library_source.GetLibSource.create_kallisto_index',
-            lambda *args, **kwargs: tmpdir / "kallisto.idx",
-        )
-
-        # Mock the get_source method to return a specific result
-        monkeypatch.setattr(
-            'htsinfer.get_library_source.GetLibSource.get_source',
-            lambda *args, **kwargs: SOURCE_FRUIT_FLY,
-        )
-
-        result = test_instance.evaluate()
-
-        assert result.file_1.taxon_id == SOURCE_FRUIT_FLY.taxon_id
-        assert result.file_1.short_name == SOURCE_FRUIT_FLY.short_name
-
-        assert result.file_2.taxon_id == SOURCE_FRUIT_FLY.taxon_id
-        assert result.file_2.short_name == SOURCE_FRUIT_FLY.short_name
-
     def test_evaluate_tax_id_not_none_name_found(self, monkeypatch, tmpdir):
         """Test when self.tax_id is not None and src_name is found."""
         CONFIG.args.tax_id = 7227
@@ -432,9 +397,9 @@ def test_evaluate_tax_id_not_none_name_found(self, monkeypatch, tmpdir):
         CONFIG.args.out_dir = tmpdir
         test_instance = GetLibSource(config=CONFIG)
 
-        # Mock the get_organism_name method to return a specific result
+        # Mock the get_source_name method to return a specific result
         monkeypatch.setattr(
-            'htsinfer.get_library_source.GetLibSource.get_organism_name',
+            'htsinfer.get_library_source.GetLibSource.get_source_name',
             lambda *args, **kwargs: "dmelanogaster",
         )