def test_section_input_process(sections, lookup):
np.random.seed(0)
d = DarwinianShift(data=MUTATION_DATA_FILE,
exon_file=EXON_FILE,
reference_fasta=REFERENCE_FASTA_FILE,
lookup=lookup,
sections=sections,
statistics=[CDFMonteCarloTest(), ChiSquareTest()],
testing_random_seed=1,
verbose=True
)
d.run_all()
if 'start' in sections.columns:
test_name = 'sections1'
else:
test_name = 'sections2'
# output new test file. Do not uncomment unless results have changed and confident new results are correct
# Be careful to only overwrite for the particular test(s) the you want.
# This function runs with many parameters. Uncommenting and running all will overwrite all cases.
# pickle.dump(d.scored_data, open(os.path.join(RESULTS_DIR, "scored_data_{}.pickle".format(test_name)), 'wb'))
# pickle.dump(d.results, open(os.path.join(RESULTS_DIR, "results_{}.pickle".format(test_name)), 'wb'))
expected = pickle.load(open(os.path.join(RESULTS_DIR, "scored_data_{}.pickle".format(test_name)), 'rb'))
assert_frame_equal(sort_dataframe(d.scored_data), sort_dataframe(expected), check_dtype=False)
expected = pickle.load(open(os.path.join(RESULTS_DIR, "results_{}.pickle".format(test_name)), 'rb'))
assert_frame_equal(sort_dataframe(d.results), sort_dataframe(expected))
def run_full_process(spectra, lookup, gene_list, transcript_list, deduplicate, excluded_positions,
use_longest_transcript_only, exclude_synonymous, exclude_nonsense):
np.random.seed(0)
# This will be a very slow test.
# Checks that the process works from start to finish for all options.
if gene_list is not None or transcript_list is not None:
# Results should be same if any of these are defined. Always testing the same transcripts
gene_list_name = '2genes'
else:
gene_list_name = 'None'
if excluded_positions is not None:
ep = '1'
else:
ep = '0'
excluded_mutations = []
if exclude_synonymous:
excluded_mutations.append('synonymous')
if exclude_nonsense:
excluded_mutations.append('nonsense')
test_name = "_".join([lookup.__class__.__name__, gene_list_name, str(int(deduplicate)), ep,
str(int(use_longest_transcript_only)), str(int(exclude_synonymous)),
str(int(exclude_nonsense))])
statistics = [CDFMonteCarloTest(num_draws=1000), ChiSquareTest(),
MonteCarloTest(stat_function=np.mean, num_draws=1000),
KSTest()]
d = DarwinianShift(data=MUTATION_DATA_FILE,
exon_file=EXON_FILE,
reference_fasta=REFERENCE_FASTA_FILE,
lookup=lookup,
statistics=statistics,
spectra=spectra['spectra'],
gene_list=gene_list,
transcript_list=transcript_list,
deduplicate=deduplicate,
excluded_positions=excluded_positions,
use_longest_transcript_only=use_longest_transcript_only,
excluded_mutation_types=excluded_mutations,
testing_random_seed=1,
verbose=True
)
d.run_all()
# output new test file. Do not uncomment unless results have changed and confident new results are correct
# Be careful to only overwrite for the particular test(s) the you want.
# This function runs with many parameters. Uncommenting and running all will overwrite all cases.
# pickle.dump(d.scored_data, open(os.path.join(RESULTS_DIR, "scored_data_{}.pickle".format(test_name)), 'wb'))
# pickle.dump(d.results, open(os.path.join(RESULTS_DIR, "results_{}.pickle".format(test_name)), 'wb'))
expected = pickle.load(open(os.path.join(RESULTS_DIR, "scored_data_{}.pickle".format(test_name)), 'rb'))
assert_frame_equal(sort_dataframe(d.scored_data), sort_dataframe(expected), check_dtype=False)
expected = pickle.load(open(os.path.join(RESULTS_DIR, "results_{}.pickle".format(test_name)), 'rb'))
assert_frame_equal(sort_dataframe(d.results), sort_dataframe(expected))
def test_incorrect_bases():
d = DarwinianShift(data=os.path.join(TEST_DATA_DIR, 'test_mutation_data_wrong_bases.tsv'),
exon_file=EXON_FILE,
reference_fasta=REFERENCE_FASTA_FILE,
lookup=lambda x: [1]*len(x.null_mutations), # Make simple lookup here
spectra=[GlobalKmerSpectrum(k=3)],
low_mem=False)
assert d.total_spectrum_ref_mismatch == 87
s1 = d.run_transcript('ENST00000263388')
assert s1.ref_mismatch_count == 77
def proj():
np.random.seed(0)
d = DarwinianShift(data=MUTATION_DATA_FILE,
exon_file=EXON_FILE,
reference_fasta=REFERENCE_FASTA_FILE,
lookup=DummyValuesRandom(random_function=np.random.random, testing_random_seed=0),
statistics=[CDFMonteCarloTest(testing_random_seed=0), ChiSquareTest()],
spectra=(EvenMutationalSpectrum(),
TranscriptKmerSpectrum(k=1),
GlobalKmerSpectrum(k=3))
)
d.run_all()
return d
def seq_pdb():
np.random.seed(0)
d = DarwinianShift(data=MUTATION_DATA_FILE,
exon_file=EXON_FILE,
reference_fasta=REFERENCE_FASTA_FILE,
lookup=DummyValuesRandom(random_function=np.random.random, testing_random_seed=0),
statistics=[CDFMonteCarloTest(testing_random_seed=0), ChiSquareTest()],
spectra=(EvenMutationalSpectrum(),
TranscriptKmerSpectrum(k=1),
GlobalKmerSpectrum(k=3)),
pdb_directory=TEST_DATA_DIR,
sifts_directory=TEST_DATA_DIR,
download_sifts=False
)
s = d.make_section(dict(transcript_id='ENST00000263388', pdb_id='4ZLP', pdb_chain='A', start=1378, end=1640))
s.apply_scores()
return s
def project():
# General project that can be used for many tests but doesn't take too long to load.
d = DarwinianShift(
data=MUTATION_DATA_FILE,
exon_file=EXON_FILE,
reference_fasta=REFERENCE_FASTA_FILE,
lookup=lambda x: [1] * len(x.null_mutations
), # Make simple lookup here
# stats=[CDFMonteCarloTest(testing_random_seed=0), ChiSquareTest()],
spectra=[GlobalKmerSpectrum(k=3)],
low_mem=False)
return d
def test_neutral():
exon_file = os.path.join(FILE_DIR, 'gene1_exons.txt')
reference_fasta = os.path.join(FILE_DIR, 'gene1.fasta')
d1 = DarwinianShift(data=os.path.join(FILE_DIR, 'data1_neutral.tsv'),
exon_file=exon_file,
reference_fasta=reference_fasta,
spectra=os.path.join(FILE_DIR, 'spectrum1.txt'))
d2 = DarwinianShift(data=os.path.join(FILE_DIR, 'data2_neutral.tsv'),
exon_file=exon_file,
reference_fasta=reference_fasta,
spectra=os.path.join(FILE_DIR, 'spectrum2.txt'))
s1 = d1.make_section(gene='gene1')
s1.load_section_mutations()
s2 = d2.make_section(gene='gene1')
s2.load_section_mutations()
res = homtest_sections(s1, s2, use_weights=True)
# output new test file. Do not uncomment unless results have changed and confident new results are correct
# pickle.dump(res, open(os.path.join(FILE_DIR, "res_neutral.pickle"), 'wb'))
expected = pickle.load(
open(os.path.join(FILE_DIR, "res_neutral.pickle"), 'rb'))
assert res == expected
def project_spectrum():
# A project that can do lots of the spectra
# Needs to be set up with the spectra in advance to the correct kmers are collected.
d = DarwinianShift(
data=MUTATION_DATA_FILE,
exon_file=EXON_FILE,
reference_fasta=REFERENCE_FASTA_FILE,
lookup=lambda x: [1] * len(x.null_mutations
), # Make simple lookup here
# stats=[CDFMonteCarloTest(testing_random_seed=0), ChiSquareTest()],
low_mem=False,
spectra=[
GlobalKmerSpectrum(
deduplicate_spectrum=False,
k=3, # Size of kmer nucleotide context. Use 3 for trinucleotides.
ignore_strand=False,
missing_value=
0, # To replace missing values. Useful to make non-zero in some cases.
name='glob_k3'),
GlobalKmerSpectrum(
deduplicate_spectrum=False,
k=1, # Size of kmer nucleotide context. Use 3 for trinucleotides.
ignore_strand=False,
missing_value=0,
# To replace missing values. Useful to make non-zero in some cases.
name='glob_k1'),
GlobalKmerSpectrum(
deduplicate_spectrum=False,
k=5,
# Size of kmer nucleotide context. Use 3 for trinucleotides.
ignore_strand=False,
missing_value=0,
# To replace missing values. Useful to make non-zero in some cases.
name='glob_k5'),
GlobalKmerSpectrum(
deduplicate_spectrum=True,
k=3,
# Size of kmer nucleotide context. Use 3 for trinucleotides.
ignore_strand=False,
missing_value=0,
# To replace missing values. Useful to make non-zero in some cases.
name='glob_k3_dd'),
GlobalKmerSpectrum(
deduplicate_spectrum=False,
k=3, # Size of kmer nucleotide context. Use 3 for trinucleotides.
ignore_strand=True,
missing_value=0,
# To replace missing values. Useful to make non-zero in some cases.
name='glob_k3_is')
])
return d