def test_nucleotide_search_unaligned_reads_output_blast_format(self):
"""
Test the unaligned reads and the store alignments
Test with a bowtie2/sam output file
Test the aligned reads file created is of the blastm8 format
"""
# create a set of alignments
alignments=store.Alignments()
unaligned_reads_store=store.Reads()
# turn off query/subject filtering
config.nucleotide_subject_coverage_threshold = 0
config.nucleotide_query_coverage_threshold = 0
config.file_basename="TEST"
# read in the aligned and unaligned reads
[unaligned_reads_file_fasta, reduced_aligned_reads_file] = nucleotide.unaligned_reads(
cfg.sam_file_annotations, alignments, unaligned_reads_store, keep_sam=True)
# reset query/subject filtering
config.nucleotide_subject_coverage_threshold = self.default_nucleotide_subject_coverage_threshold
config.nucleotide_query_coverage_threshold = self.default_nucleotide_query_coverage_threshold
# test file is of the blastm8 format
file_format=utilities.determine_file_format(reduced_aligned_reads_file)
# remove temp files
utils.remove_temp_file(unaligned_reads_file_fasta)
utils.remove_temp_file(reduced_aligned_reads_file)
self.assertEqual(file_format,"blastm8")
def test_determine_file_format_fastq_gzip(self):
"""
Test the determine_file_format function with a fastq file that is gzipped
"""
# create a small gzipped fastq file
# read in the small fastq file
file_handle = open(cfg.small_fastq_file, "rt")
# create a temp file
file_out, gzip_fastq_file = tempfile.mkstemp(suffix=".gz")
os.close(file_out)
# write the gzipped file
file_handle_gzip = gzip.open(gzip_fastq_file, "wt")
shutil.copyfileobj(file_handle, file_handle_gzip)
file_handle.close()
file_handle_gzip.close()
format = utilities.determine_file_format(gzip_fastq_file)
# remove the temp gzipped file
utils.remove_temp_file(gzip_fastq_file)
self.assertEqual(format, "fastq.gz")
def test_nucleotide_search_unaligned_reads_output_fasta_format(self):
"""
Test the unaligned reads and the store alignments
Test with a bowtie2/sam output file
Test output file is of fasta format
Test sam file is not removed
"""
# create a set of alignments
alignments=store.Alignments()
unaligned_reads_store=store.Reads()
# turn off query/subject filtering
config.nucleotide_subject_coverage_threshold = 0
config.nucleotide_query_coverage_threshold = 0
# read in the aligned and unaligned reads
[unaligned_reads_file_fasta, reduced_aligned_reads_file] = nucleotide.unaligned_reads(
cfg.sam_file_unaligned_reads, alignments, unaligned_reads_store, keep_sam=True)
# reset query/subject filtering
config.nucleotide_subject_coverage_threshold = self.default_nucleotide_subject_coverage_threshold
config.nucleotide_query_coverage_threshold = self.default_nucleotide_query_coverage_threshold
# check for fasta output file format
file_format=utilities.determine_file_format(unaligned_reads_file_fasta)
self.assertEqual("fasta",file_format)
# remove temp files
utils.remove_temp_file(unaligned_reads_file_fasta)
utils.remove_temp_file(reduced_aligned_reads_file)
def test_determine_file_format_genetable(self):
"""
Test the determine_file_format function with a gene table tsv file
"""
format = utilities.determine_file_format(cfg.genetable_file)
self.assertEqual(format, "genetable")
def test_determine_file_format_biom(self):
"""
Test the determine_file_format function with a biom file
"""
format = utilities.determine_file_format(cfg.biom_file)
self.assertEqual(format, "biom")
def test_determine_file_format_sam_without_header(self):
"""
Test the determine_file_format function with a sam file without header
"""
format = utilities.determine_file_format(cfg.sam_file_without_header)
self.assertEqual(format, "sam")
def test_determine_file_format_fastq(self):
"""
Test the determine_file_format function with a fastq file
"""
format = utilities.determine_file_format(cfg.small_fastq_file)
self.assertEqual(format, "fastq")
def test_determine_file_format_rapsearch2_without_header(self):
"""
Test the determine_file_format function with a rapsearch2 output file
without the standard header format
"""
format = utilities.determine_file_format(
cfg.rapsearch2_output_file_without_header)
self.assertEqual(format, "blastm8")