Added unit tests for pubMLST MLST profiling

src/automlst/engine/local/abif.py

@@ -1,7 +1,7 @@
 import asyncio
 from numbers import Number
 from os import path
-from typing import AsyncGenerator, Collection, Sequence, Union
+from typing import Any, AsyncGenerator, Collection, Sequence, Union
 from automlst.engine.data.genomics import NamedString, SangerTraceData
 from Bio.SeqRecord import SeqRecord
 from Bio import SeqIO, Align
@@ -104,12 +104,15 @@ async def read_abif(seq_path: str) -> SangerTraceData:
     )
     return trace_data
 
+
 def _biopython_local_pairwise_alignment(reference: NamedString, query: NamedString) -> tuple[NamedString, NamedString]:
     aligner = Align.PairwiseAligner(scoring="blastn")
     aligner.mode = "local"
-    alignment_result = sorted(aligner.align(reference.sequence, query.sequence))[0] # take the best alignment
+    alignment_result = sorted(aligner.align(reference.sequence, query.sequence))[
+        0]  # take the best alignment
     return NamedString(alignment_result.sequences[0].id, alignment_result.sequences[0].seq), NamedString(alignment_result.sequences[1].id, alignment_result.sequences[1].seq)
 
+
 async def reference_consensus_assembly(reference: NamedString, sanger_traces: Collection[SangerTraceData]) -> AsyncGenerator[NamedString, Any]:
     for sanger_trace in sanger_traces:
         yield (await asyncio.to_thread(_biopython_local_pairwise_alignment, reference, sanger_trace))[1]

src/automlst/engine/remote/databases/institutpasteur/mlst.py

@@ -30,15 +30,11 @@ class InstitutPasteurProfiler(MLSTProfiler):
                 alelle_id = allele["allele_id"]
                 yield Allele(allele_loci=allele_loci, allele_variant=alelle_id)
 
-    async def fetch_mlst_st(self, schema_id: int, alleles: Union[AsyncIterable[Allele], Iterable[Allele]]) -> MLSTProfile:
+    async def fetch_mlst_st(self, schema_id: int, alleles: AsyncIterable[Allele]) -> MLSTProfile:
         uri_path = f"schemes/{schema_id}/designations"
         allele_request_dict: dict[str, list[dict[str, str]]] = defaultdict(list)
-        if isinstance(alleles, AsyncIterable):
         async for allele in alleles:
             allele_request_dict[allele.allele_loci].append({"allele": str(allele.allele_variant)})
-        else:
-            for allele in alleles:
-                allele_request_dict[allele.allele_loci].append({"allele": str(allele.allele_variant)})
         response = await self._http_client.post(uri_path, json={
             "designations": allele_request_dict
         })

src/automlst/engine/remote/databases/mlst.py

@@ -1,6 +1,6 @@
 from abc import abstractmethod
 from contextlib import AbstractAsyncContextManager
-from typing import AsyncGenerator, AsyncIterable, Generator, Iterable, Mapping, Union
+from typing import Any, AsyncGenerator, AsyncIterable, Generator, Iterable, Mapping, Union
 
 from aiohttp import ClientSession
 

tests/nsbdiagnosistoolkit/engine/remote/databases/institutpasteur/test_mlst.py

@@ -16,6 +16,7 @@ async def test_profiling_results_in_exact_matches_when_exact():
         assert len(targets_left) == 0
 
 async def test_profiling_results_in_correct_st():
+    async def dummy_allele_generator():
         dummy_alleles = [
         Allele("adk", "1"),
         Allele("fumC", "1"),
@@ -25,8 +26,10 @@ async def test_profiling_results_in_correct_st():
         Allele("pepA", "1"),
         Allele("pgm", "1"),
         ]
+        for dummy_allele in dummy_alleles:
+            yield dummy_allele
     async with InstitutPasteurProfiler(database_name="pubmlst_bordetella_seqdef") as dummy_profiler:
-        mlst_st_data = await dummy_profiler.fetch_mlst_st(3, dummy_alleles)
+        mlst_st_data = await dummy_profiler.fetch_mlst_st(3, dummy_allele_generator())
         assert mlst_st_data is not None
         assert isinstance(mlst_st_data, MLSTProfile)
         assert mlst_st_data.clonal_complex == "ST-2 complex"

tests/nsbdiagnosistoolkit/engine/remote/databases/pubmlst/test_mlst.py

@@ -1,6 +1,6 @@
+import asyncio
 from Bio import SeqIO
 from automlst.engine.data.mlst import Allele, MLSTProfile
-from automlst.engine.remote.databases.institutpasteur.mlst import InstitutPasteurProfiler
 from automlst.engine.remote.databases.pubmlst.mlst import PubMLSTProfiler
 
 
@@ -24,6 +24,7 @@ async def test_profiling_results_in_exact_matches_when_exact():
         assert len(dummy_alleles) == 0
 
 async def test_profiling_results_in_correct_st():
+    async def generate_dummy_targets():
         dummy_alleles = [
                 Allele("adk", "1"),
                 Allele("atpG", "1"),
@@ -33,16 +34,18 @@ async def test_profiling_results_in_correct_st():
                 Allele("pgi", "1"),
                 Allele("recA", "5"),
             ]
-    async with InstitutPasteurProfiler(database_name="pubmlst_hinfluenzae_seqdef") as dummy_profiler:
+        for dummy_allele in dummy_alleles:
-        mlst_st_data = await dummy_profiler.fetch_mlst_st(1, dummy_alleles)
+            yield dummy_allele
+    async with PubMLSTProfiler(database_name="pubmlst_hinfluenzae_seqdef") as dummy_profiler:
+        mlst_st_data = await dummy_profiler.fetch_mlst_st(1, generate_dummy_targets())
         assert mlst_st_data is not None
         assert isinstance(mlst_st_data, MLSTProfile)
         assert mlst_st_data.clonal_complex == "ST-3 complex"
         assert mlst_st_data.sequence_type == "3"
 
 async def test_sequence_profiling_is_correct():
-    sequence = str(SeqIO.read("tests/resources/tohama_I_bpertussis.fasta", "fasta").seq)
+    sequence = str(SeqIO.read("tests/resources/FDAARGOS_1560.fasta", "fasta").seq)
-    async with InstitutPasteurProfiler(database_name="pubmlst_hinfluenzae_seqdef") as dummy_profiler:
+    async with PubMLSTProfiler(database_name="pubmlst_hinfluenzae_seqdef") as dummy_profiler:
         profile = await dummy_profiler.profile_string(1, sequence)
         assert profile is not None
         assert isinstance(profile, MLSTProfile)