def test_create_fastq_dict(
valid_files_with_index,
valid_files_with_index_dict,
invalid_files_one_lane_indexed,
invalid_files_one_lane_indexed_dict,
):
fastq_dict = tenx_utils.create_fastq_dict(valid_files_with_index)
assert fastq_dict == valid_files_with_index_dict
fastq_dict = tenx_utils.create_fastq_dict(invalid_files_one_lane_indexed)
assert fastq_dict == invalid_files_one_lane_indexed_dict
def test_get_fastqs_for_read_index(valid_files_with_index):
fastq_dict = tenx_utils.create_fastq_dict(valid_files_with_index)
fastqs = tenx_utils.get_fastqs_for_read_index(fastq_dict, 'read1')
assert fastqs == ['gs://5/r1.fastq.gz', 'gs://7/r1.fastq.gz']
fastqs = tenx_utils.get_fastqs_for_read_index(fastq_dict, 'read2')
assert fastqs == ['gs://5/r2.fastq.gz', 'gs://7/r2.fastq.gz']
fastqs = tenx_utils.get_fastqs_for_read_index(fastq_dict, 'index1')
assert fastqs == ['gs://5/i1.fastq.gz', 'gs://7/i1.fastq.gz']
def get_optimus_inputs(primary_bundle):
"""Gather the necessary inputs for Optimus from the bundle metadata object.
Args:
humancellatlas.data.metadata.Bundle (obj): A bundle metadata object.
Returns:
tuple: tuple of the sample_id, ncbi_taxon_id, dict mapping flow cell lane indices
to fastq file manifests
Raises:
requests.HTTPError: on 4xx errors or 5xx errors beyond the timeout
"""
sample_id = metadata_utils.get_sample_id(primary_bundle)
ncbi_taxon_id = metadata_utils.get_ncbi_taxon_id(primary_bundle)
fastq_files = primary_bundle.sequencing_output
lane_to_fastqs = tenx_utils.create_fastq_dict(fastq_files)
library_construction_method = metadata_utils.get_library_construction_method_ontology(
primary_bundle)
chemistry = get_tenx_chemistry(library_construction_method)
return sample_id, ncbi_taxon_id, lane_to_fastqs, chemistry
def test_create_fastq_dict_raises_error_when(
missing_lane_index_multiple_sets_of_reads):
with pytest.raises(tenx_utils.InsufficientLaneInfoError):
tenx_utils.create_fastq_dict(missing_lane_index_multiple_sets_of_reads)
def test_create_fastq_dict_reindexes_with_zero_when(
missing_lane_index_one_set_of_reads):
fastq_dict = tenx_utils.create_fastq_dict(
missing_lane_index_one_set_of_reads)
assert 0 in fastq_dict
assert None not in fastq_dict
def test_validate_lanes_accepts_lanes_when_none_indexed(valid_files_no_index):
fastq_dict = tenx_utils.create_fastq_dict(valid_files_no_index)
tenx_utils.validate_lanes(fastq_dict)
def test_validate_lanes_rejects_mixing_indexed_and_non_indexed_lanes(
invalid_files_one_lane_indexed):
fastq_dict = tenx_utils.create_fastq_dict(invalid_files_one_lane_indexed)
with pytest.raises(tenx_utils.LaneMissingFileError):
tenx_utils.validate_lanes(fastq_dict)
def test_validate_lanes_requires_read2(invalid_files_missing_read2):
fastq_dict = tenx_utils.create_fastq_dict(invalid_files_missing_read2)
with pytest.raises(tenx_utils.LaneMissingFileError):
tenx_utils.validate_lanes(fastq_dict)