Completed CLI for B. pertussis MLST typing from fasta files

This commit is contained in:
2025-01-03 19:20:19 +00:00
parent 362e0867e5
commit e9977d2da7
17 changed files with 230 additions and 51 deletions

View File

@@ -1,8 +1,8 @@
import os
from mlstmyfasta.engine.local.abif import load_sanger_sequence
from mlstmyfasta.engine.local.abif import read_abif
async def test_load_sanger_sequence_has_data():
assert os.path.exists("tests/resources/1I1_F_P1815443_047.ab1")
result_data = await load_sanger_sequence("tests/resources/1I1_F_P1815443_047.ab1")
result_data = await read_abif("tests/resources/1I1_F_P1815443_047.ab1")
assert result_data is not None

View File

@@ -1,16 +0,0 @@
from Bio import SeqIO
from mlstmyfasta.engine.data.MLST import Allele
from mlstmyfasta.engine.remote.databases.institutpasteur.profiling import InstitutPasteurProfiler
async def test_profiling_results_in_exact_matches_when_exact():
sequence = str(SeqIO.read("tests/resources/tohama_I_bpertussis.fasta", "fasta").seq)
async with InstitutPasteurProfiler(database_name="pubmlst_bordetella_seqdef") as dummy_profiler:
exact_matches = dummy_profiler.fetch_mlst_profile(sequence_string=sequence)
targets_left = {"adk", "fumC", "glyA", "tyrB", "icd", "pepA", "pgm"}
async for exact_match in exact_matches:
assert isinstance(exact_match, Allele)
assert exact_match.allele_variant == '1' # All of Tohama I has allele id I
targets_left.remove(exact_match.allele_loci)
assert len(targets_left) == 0

View File

@@ -0,0 +1,35 @@
from Bio import SeqIO
from mlstmyfasta.engine.data.MLST import Allele, MLSTProfile
from mlstmyfasta.engine.remote.databases.institutpasteur.profiling import InstitutPasteurProfiler
async def test_profiling_results_in_exact_matches_when_exact():
sequence = str(SeqIO.read("tests/resources/tohama_I_bpertussis.fasta", "fasta").seq)
async with InstitutPasteurProfiler(database_name="pubmlst_bordetella_seqdef") as dummy_profiler:
exact_matches = dummy_profiler.fetch_mlst_allele_variants(sequence_string=sequence)
targets_left = {"adk", "fumC", "glyA", "tyrB", "icd", "pepA", "pgm"}
async for exact_match in exact_matches:
assert isinstance(exact_match, Allele)
assert exact_match.allele_variant == '1' # All of Tohama I has allele id I
targets_left.remove(exact_match.allele_loci)
assert len(targets_left) == 0
async def test_profiling_results_in_correct_st():
sequence = str(SeqIO.read("tests/resources/tohama_I_bpertussis.fasta", "fasta").seq)
dummy_alleles = [
Allele("adk", "1"),
Allele("fumC", "1"),
Allele("glyA", "1"),
Allele("tyrB", "1"),
Allele("icd", "1"),
Allele("pepA", "1"),
Allele("pgm", "1"),
]
async with InstitutPasteurProfiler(database_name="pubmlst_bordetella_seqdef") as dummy_profiler:
exact_matches = dummy_profiler.fetch_mlst_allele_variants(sequence_string=sequence)
mlst_st_data = await dummy_profiler.fetch_mlst_st(dummy_alleles)
assert mlst_st_data is not None
assert isinstance(mlst_st_data, MLSTProfile)
assert mlst_st_data.clonal_complex == "ST-2 complex"
assert mlst_st_data.sequence_type == "1"