def test_non_coding_exons_gprin1(gprin1):
trx_data = transcript_info.read_transcript_info(gprin1['tsl'],
gprin1['exontable'],
gprin1['seqs'],
remove_na=False)
# The last two exons of ENSMUST00000099506 are non-coding.
# The unique coding exon has UTR at both ends.
assert ''.join(
str(exon) for exon in trx_data.loc[
trx_data['TranscriptIDCluster'].
map(lambda ids: 'ENSMUST00000099506' in ids.split('/')),
'ExonProteinSequence']) == (
'MRDCCSSPKAIPAPPRHALDQSLGMDPRHTSSSGAAEGASCSERPAGSLACPSPNCSPLP'
'ETPRAHGALTSDNSGTTLFGKPEPMSSAEATPTASEIRNPVFSGKMDGNSLKQADSTSTR'
'KEEAGSLRNEESMLKGKAEPMIYGKGEPGTVGRVDCTASGAENSGSLGKVDMPCSSKVDI'
'VSPGGDNAGSLRKVETISSGKMDPKTENVMHSRRERPGSTGEGDLVSLRENDMKPPDNTD'
'SASTKKTDPEFSGKLTPGSSGKTELVSSVTVAPVTSENVNPVCSGGAGPAAVGNSETLSS'
'VKKDPQLLGKKEAVSSGEGGSVSVRMAETVSARQPEGMFPAKTDSTSSNSTGPSGRADPV'
'SLRNSELVSPVKPERLSSGQAERVSLVKTETLSSGKEDPRSSRRVDHTTVTGNMQTSQKG'
'NPESSGKTDLGSSSSGDTRSLGTWGSLSAAKAEVTEGKGDPQPWKKASLPASEKTDPLAS'
'SKAGSASQGKAETVSPGEVDAMTLGKTVPTSSGKTALVSPGKVDLMTSERAEGIPELQAS'
'EKGNPVNSTRVDTGATGSTEPKSGVKVITQIPGATSPGKVETPSLQKEQPQLSEKTDPSR'
'KVDPPTTVEPVSLGKADSASPSPRKAESQTSAKTVPQAPDKATSSLRQSDGTPYSSAQPQ'
'RDTRSIGSLPEREPSASTSQKDLAAAAAQKSPSAEAAAPPPGPRTRDNFTKAPSWDAGAP'
'PPREDAGTQAGAQACVSVAVSPMSPQDGAGGPAFSFQAAPRAPSPAPRPPSRRDAGLQVS'
'LGAAETRSVATGPMTPQAAAPPAVPPVFPEVRVRPGSVLAAALAPQEATEPVRDVSWDEK'
'GMTWEVYGASMEVEVLGMAIQKHLERQIEEHGRQGAPAPAPPPAVRAGPGRAGSVRTAPA'
'EGAAKRPPGLFRALLQSVRRPRCCSRAGPTAE*')
def test_keep_badquality_sequences(mapk8):
trx_data = transcript_info.read_transcript_info(mapk8['tsl'],
mapk8['exontable'],
mapk8['seqs'],
remove_na=False,
remove_badquality=False)
# ENSRNOT00000083933 has Xs in its sequence: ...VILGMGYKENGQXVXHVQRGLICC*
assert sum(trx_data['TranscriptID'] == 'ENSRNOT00000083933') == 5
def get_transcripts(input_folder, max_tsl_level=3.0, species_list=None):
"""Return a DataFrame with the transcript information."""
ensembl_info = os.path.join(input_folder, 'Ensembl')
return transcript_info.read_transcript_info(
os.path.join(ensembl_info, 'tsl.csv'),
os.path.join(ensembl_info, 'exonstable.tsv'),
os.path.join(ensembl_info, 'sequences.fasta'),
max_tsl_level=max_tsl_level,
remove_na=False,
species_list=species_list)
def test_species_list(mapk8):
species_list = ['bos_taurus', 'homo_sapiens']
trx_data = transcript_info.read_transcript_info(mapk8['tsl'],
mapk8['exontable'],
mapk8['seqs'],
remove_na=False,
species_list=species_list)
assert sorted(trx_data.Species.unique()) == species_list
def test_read_transcript_info(mapk8):
trx_data = transcript_info.read_transcript_info(mapk8['tsl'],
mapk8['exontable'],
mapk8['seqs'])
assert trx_data.loc[
trx_data['TranscriptID'] == 'ENST00000374179',
'TSL'].unique()[0] == 1.0 # '1 (assigned to previous version 7)'
# Only h. sapiens & m. musculus have TSL information:
assert all(trx_data.Species.unique() == ['homo_sapiens', 'mus_musculus'])
def _get_subexon_data(folder):
test_dir = os.path.dirname(filename)
datadir = os.path.join(test_dir, 'data')
folder_path = os.path.join(datadir, folder, 'Ensembl')
trx_data = transcript_info.read_transcript_info(
os.path.join(folder_path, 'tsl.csv'),
os.path.join(folder_path, 'exonstable.tsv'),
os.path.join(folder_path, 'sequences.fasta'),
remove_na=False)
clustered = transcript_info.exon_clustering(trx_data)
return subexons.create_subexon_table(clustered)
def _get_clustered_trx_data(folder):
test_dir = os.path.dirname(filename)
datadir = os.path.join(test_dir, 'data')
folder_path = os.path.join(datadir, folder, 'Ensembl')
species_list = None
if folder == "MAPK8_all":
species_list = ['anser_brachyrhynchus']
trx_data = transcript_info.read_transcript_info(
os.path.join(folder_path, 'tsl.csv'),
os.path.join(folder_path, 'exonstable.tsv'),
os.path.join(folder_path, 'sequences.fasta'),
remove_na=False,
species_list=species_list)
return transcript_info.exon_clustering(trx_data)
def test_non_coding_exons_camk2a(camk2a):
trx_data = transcript_info.read_transcript_info(camk2a['tsl'],
camk2a['exontable'],
camk2a['seqs'],
remove_na=False)
# The two first exons of ENSSSCT00000052397 are non-coding
assert ''.join(
str(exon) for exon in trx_data.loc[
trx_data['TranscriptID'] == 'ENSSSCT00000052397',
'ExonProteinSequence']) == (
'MLLFLALWALVPCLVLLSLYFYSSAGGKSGGNKKNDGVKKRKSSSSVQLMESSESTNTTI'
'EDEDTKVRKQEIIKVTEQLIEAISNGDFESYTKMCDPGMTAFEPEALGNLVEGLDFHRFY'
'FENLWSRNSKPVHTTILNPHIHLMGDESACIAYIRITQYLDAGGIPRTAQSEETRVWHRR'
'DGKWQIVHFHRSGAPSVLPH*')
def test_remove_na(mapk8):
trx_data = transcript_info.read_transcript_info(mapk8['tsl'],
mapk8['exontable'],
mapk8['seqs'],
remove_na=False)
# I keep other species, not only h. sapiens & m. musculus:
assert len(trx_data.Species.unique()) > 2
# remove_na doesn't interfere with selecting the correct biotype
assert sum(value not in ['Protein coding', 'protein_coding']
for value in trx_data.Biotype.unique()) == 0
# ENSRNOT00000083933 has Xs in its sequence: ...VILGMGYKENGQXVXHVQRGLICC*
assert sum(trx_data['TranscriptID'] == 'ENSRNOT00000083933') == 0
def test_exon_clustering(mapk8):
trx_data = transcript_info.read_transcript_info(mapk8['tsl'],
mapk8['exontable'],
mapk8['seqs'])
clustered = transcript_info.exon_clustering(trx_data)
# Input order doesn't change
assert all(clustered['ExonID'] == trx_data['ExonID'])
# Exon can not have more than one cluster
assert all(
clustered.groupby('ExonID').apply(
lambda df: len(df['Cluster'].unique()) == 1))
# Non-clustered exons have Cluster == 0 and QueryExon == ''
assert all(row.QueryExon == '' for row in clustered.itertuples()
if row.Cluster == 0)
assert all(row.Cluster == 0 for row in clustered.itertuples()
if row.QueryExon == '')
# Sequences with less than 4 residues are non-clustered by default
assert all(clustered.loc[clustered['ExonProteinSequence'].map(len) < 4,
'Cluster'] == 0)
for _, group in clustered.groupby('Cluster'):
nans = np.isnan(group['PercentIdentity'])
# There is a nan in PercentIdentity when a sequence initialize its own
# cluster, so the QueryExon and the ExonID should be the same
assert np.all(group.loc[nans, 'QueryExon'] == group.loc[nans,
'ExonID'])
# Also, if there are more exons, the exons with nan should be the
# QueryExon of other exon in the cluster. It can not be alone.
if len(group) > 1:
assert np.all([
exon in group.loc[np.logical_not(nans), 'QueryExon'].values
for exon in group.loc[nans, 'QueryExon'].unique()
])
# The aligned seq should be in the exon
subset = group[np.logical_not(nans)]
for _, exon in subset.groupby('ExonIDCluster'):
assert np.any([((row.AlignedTarget.replace('-', '')
in row.ExonProteinSequence)
or (row.AlignedQuery.replace(
'-', '')) in row.ExonProteinSequence)
for row in exon.itertuples()])