def test_chromosome_names():
test_fasta = reftools.RefFasta(os.path.join(dir, "toy.fasta"),
"samtools",
"bwa",
no_initial_index=True)
names = test_fasta.chromosome_names()
assert type(names) == tuple
assert names == ("seq1", "seq2")
def test_chromosome_lengths():
test_fasta = reftools.RefFasta(os.path.join(dir, "toy.fasta"),
"samtools",
"bwa",
no_initial_index=True)
lengths = test_fasta.chromosome_lengths()
assert type(lengths) == tuple
assert lengths == (20, 40)
def test_RefFasta():
test_fasta = reftools.RefFasta(os.path.join(dir, "toy.fasta"),
"samtools",
"bwa",
no_initial_index=True)
test_fasta2 = reftools.RefFasta(os.path.join(dir, "toy2.fasta"),
"samtools",
"bwa",
no_initial_index=True)
os.utime(os.path.join(dir, "toy.dict"), None)
assert test_fasta.filepath == os.path.join(dir, "toy.fasta")
assert test_fasta.is_faidxed() is True
assert test_fasta.check_bwa_index() is True
assert test_fasta.check_seq_dict() is True
assert test_fasta2.filepath == os.path.join(dir, "toy2.fasta")
assert test_fasta2.is_faidxed() is False
assert test_fasta2.check_bwa_index() is False
assert test_fasta2.check_seq_dict() is False
def test_check_bam_fasta_compatibility():
test_bam = bam.BamFile(os.path.join(dir, "toy.bam"),
"samtools",
no_initial_index=True)
test2_bam = bam.BamFile(os.path.join(dir, "tinyheader.bam"),
"samtools",
no_initial_index=True)
test_fasta = reftools.RefFasta(os.path.join(dir, "toy.fasta"),
"samtools",
"bwa",
no_initial_index=True)
test2_fasta = reftools.RefFasta(os.path.join(dir, "toy3.fasta"),
"samtools",
"bwa",
no_initial_index=True)
assert utils.check_bam_fasta_compatibility(test_bam, test_fasta) is True
assert utils.check_bam_fasta_compatibility(test2_bam, test_fasta) is False
assert utils.check_bam_fasta_compatibility(test_bam, test2_fasta) is False
def test_get_chrom_length():
test_fasta = reftools.RefFasta(os.path.join(dir, "toy.fasta"),
"samtools",
"bwa",
no_initial_index=True)
lengths = test_fasta.get_chrom_length("seq1")
assert type(lengths) == int
assert lengths == 20
with pytest.raises(RuntimeError):
lengths = test_fasta.get_chrom_length("foo")
def test_mask_reference():
test_fasta = reftools.RefFasta(os.path.join(dir, "toy.fasta"),
"samtools",
"bwa",
no_initial_index=True)
new_path = test_fasta.mask_reference(os.path.join(dir, "toy_mask.bed"),
os.path.join(dir, "toy.masked.fasta"))
assert new_path == os.path.join(dir, "toy.masked.fasta")
assert os.path.exists(os.path.join(dir, "toy.masked.fasta.fai"))
contents = read_file_to_list(os.path.join(dir, "toy.masked.fasta"))
assert contents[1] == "NNNNNNNNNNNNNNNNNNNN"
def test_RefFasta_setup():
test_fasta2 = reftools.RefFasta(os.path.join(dir, "toy2.fasta"),
"samtools",
"bwa",
no_initial_index=False)
test_fasta2.conditional_index_bwa()
test_fasta2.conditional_seq_dict()
assert os.path.exists(os.path.join(dir, "toy2.fasta.amb"))
assert os.path.exists(os.path.join(dir, "toy2.fasta.ann"))
assert os.path.exists(os.path.join(dir, "toy2.fasta.bwt"))
assert os.path.exists(os.path.join(dir, "toy2.fasta.fai"))
assert os.path.exists(os.path.join(dir, "toy2.fasta.pac"))
assert os.path.exists(os.path.join(dir, "toy2.fasta.sa"))
assert os.path.exists(os.path.join(dir, "toy2.fasta.dict"))
def test_chromosome_bed():
test_fasta = reftools.RefFasta(os.path.join(dir, "toy.fasta"),
"samtools",
"bwa",
no_initial_index=True)
path1 = test_fasta.chromosome_bed(os.path.join(dir, "reftest1.bed"),
["seq1"])
path2 = test_fasta.chromosome_bed(os.path.join(dir, "reftest2.bed"),
["seq1", "seq2"])
assert path1 == os.path.join(dir, "reftest1.bed")
assert path2 == os.path.join(dir, "reftest2.bed")
assert read_bed(os.path.join(dir, "reftest1.bed")) == [["seq1", "0", "20"]]
assert read_bed(os.path.join(dir, "reftest2.bed")) == [["seq1", "0", "20"],
["seq2", "0", "40"]]
with pytest.raises(RuntimeError):
path3 = test_fasta.chromosome_bed(os.path.join(dir, "reftest1.bed"),
["foo"])
def test_isolate_chroms():
test_fasta = reftools.RefFasta(os.path.join(dir, "toy.fasta"),
"samtools",
"bwa",
no_initial_index=True)
seq1 = test_fasta.isolate_chroms(os.path.join(dir, "newfasta"), ["seq1"],
bed_mask=None)
seq1_masked = test_fasta.isolate_chroms(os.path.join(dir, "newfasta"),
["seq1"],
bed_mask=os.path.join(
dir, "toy_mask.bed"))
assert seq1 == os.path.join(dir, "newfasta.fa")
assert seq1_masked == os.path.join(dir, "newfasta.masked.fa")
assert os.path.exists(os.path.join(dir, "newfasta.fa.fai"))
assert os.path.exists(os.path.join(dir, "newfasta.masked.fa.fai"))
contents1 = read_file_to_list(os.path.join(dir, "newfasta.fa"))
contents2 = read_file_to_list(os.path.join(dir, "newfasta.masked.fa"))
assert contents1[1] == "AAAATTTTAAAATTTTGGGG"
assert contents2[1] == "NNNNNNNNNNNNNNNNNNNN"
def test_bwa_mem_mapping_sambamba():
# Test single end, no rg
toy_fasta = reftools.RefFasta(os.path.join(dir, "toy.fasta"))
test_bam = assemble.bwa_mem_mapping_sambamba(
"bwa",
"samtools",
"sambamba",
toy_fasta,
os.path.join(dir, "assemble_test"), [os.path.join(dir, "toy_1.fastq")],
1,
"None", [""],
cram=False)
assert os.path.exists(os.path.join(dir, "assemble_test.sorted.bam"))
assert pysam.idxstats(os.path.join(dir, "assemble_test.sorted.bam")
) == 'seq1\t20\t0\t0\nseq2\t40\t0\t0\n*\t0\t0\t3\n'
# Test single end, with rg
toy_fasta = reftools.RefFasta(os.path.join(dir, "toy.fasta"))
test_bam = assemble.bwa_mem_mapping_sambamba(
"bwa",
"samtools",
"sambamba",
toy_fasta,
os.path.join(dir,
"test_assemble.rg"), [os.path.join(dir, "toy_1.fastq")],
1,
"@RG\tID:{}".format("test"), [""],
cram=False)
assert os.path.exists(os.path.join(dir, "test_assemble.rg.sorted.bam"))
assert pysam.idxstats(os.path.join(dir, "test_assemble.rg.sorted.bam")
) == 'seq1\t20\t0\t0\nseq2\t40\t0\t0\n*\t0\t0\t3\n'
header = pysam.view("-H", os.path.join(dir, "test_assemble.rg.sorted.bam"))
assert header.find("@RG\\tID:test") != -1
# Test raising error for inaccessible reference
with pytest.raises(RuntimeError):
fake_fasta = reftools.RefFasta(os.path.join(dir,
"DOES_NOT_EXIST.fasta"),
no_initial_index=True)
test_bam = assemble.bwa_mem_mapping_sambamba(
"bwa",
"samtools",
"sambamba",
fake_fasta,
os.path.join(dir,
"test_assemble"), [os.path.join(dir, "toy_1.fastq")],
1,
"None", [""],
cram=False)
# Test raising error for no fastqs
toy_fasta = reftools.RefFasta(os.path.join(dir, "toy.fasta"))
with pytest.raises(RuntimeError):
test_bam = assemble.bwa_mem_mapping_sambamba("bwa",
"samtools",
"sambamba",
toy_fasta,
os.path.join(
dir, "test_assemble"),
[],
1,
"None", [""],
cram=False)
# Test raising error for inaccessible fastq
with pytest.raises(RuntimeError):
toy_fasta = reftools.RefFasta(os.path.join(dir, "toy.fasta"))
test_bam = assemble.bwa_mem_mapping_sambamba(
"bwa",
"samtools",
"sambamba",
toy_fasta,
os.path.join(dir, "test_assemble"),
[os.path.join(dir, "toy_FOO_fake.fastq")],
1,
"None", [""],
cram=False)