def ref_sequence_abundances():
from samsum import commands
from samsum import testing_utils as utils
test_sam = utils.get_test_data("samsum_test_2.sam")
test_aln = utils.get_test_data("samsum_test_2.fasta")
ref_seq_abunds = commands.ref_sequence_abundances(aln_file=test_sam,
seq_file=test_aln,
min_aln=10,
p_cov=0,
map_qual=0)
return ref_seq_abunds
def test_ref_sequence_abundances():
from samsum import commands
from samsum.classy import RefSequence
test_sam = utils.get_test_data("samsum_test_2.sam")
test_aln = utils.get_test_data("samsum_test_2.fasta")
ref_seq_abunds = commands.ref_sequence_abundances(aln_file=test_sam,
seq_file=test_aln,
min_aln=10,
p_cov=0,
map_qual=0)
e10 = ref_seq_abunds[
'AB-755_P22_E10_NODE_6_length_36342_cov_2156.57_ID_21'] # type: RefSequence
assert e10.reads_mapped == 10
assert e10.weight_total == 5.0
return
def test_get_mapped_reads():
from samsum import _sam_module
test_sam = utils.get_test_data('pytest_1.sam')
assert isinstance(test_sam, str)
mapping_list = _sam_module.get_mapped_reads(test_sam, False, 10, 0, 'q')
assert (len(mapping_list) == 16)
return
def test_ref_sequence_length():
"""
Ensure the RefSequence.rightmost doesn't exceed its length from alignment_dat_example.load_sam()
:return:
"""
from samsum import file_parsers as ss_fp
from samsum import alignment_utils as ss_aln_utils
test_seq = utils.get_test_data("samsum_test_2.fasta")
test_aln = utils.get_test_data("samsum_test_2.sam")
ref_seq_lengths = ss_fp.fasta_seq_lengths_ext(fasta_file=test_seq)
assert len(ref_seq_lengths) == 277
# Parse the alignments and return the strings of reads mapped to each reference sequence
mapped_dict = ss_fp.sam_parser_ext(test_aln, True)
# Convert the alignment strings returned by the sam_parser_ext into ss_class.AlignmentDat instances
alignments, num_unmapped, mapped_weight_sum = ss_aln_utils.load_alignments(
mapped_dict, 10)
for aln_dat in alignments: # type: samsum.classy.AlignmentDat
assert aln_dat.end < ref_seq_lengths[aln_dat.ref]
return
def test_fasta_reader():
from samsum import file_parsers as ss_fp
test_seq = utils.get_test_data("fasta_test.fa")
ref_seq_lengths = ss_fp.fasta_seq_lengths_ext(fasta_file=test_seq)
assert ref_seq_lengths == {"contig 1": 16, "contig 2": 25}
return
def sam_list():
data_path = utils.get_test_data("test-data")
filenames = sorted(glob.glob(os.path.join(data_path, "*.sam")))
return filenames