def _run_hhblits(proteins, name):
# Generates multiple sequence alignment using hhblits
sequences = [_get_seq(p) for p in proteins]
target_hhblits_path = get_target_hhblits_path(name)
check_path(target_hhblits_path)
fname = os.path.join(target_hhblits_path, name + '_seed.fasta')
# SeqIO.write(sequences, query, "fasta")
run_clustalo(sequences, fname)
output_hhblits = os.path.join(target_hhblits_path, name + '.a3m')
output_reformat1 = os.path.join(target_hhblits_path, name + '.a2m')
output_reformat2 = os.path.join(target_hhblits_path, name + '_.fasta')
db_hh = '/vol/sci/bio/data/or.zuk/projects/ContactMaps/data/uniref30/UniRef30_2020_06'
hhblits_params = '-n 3 -e 1e-3 -maxfilt 10000000000 -neffmax 20 -nodiff -realign_max 10000000000'
hhblits_cmd = f'hhblits -i {fname} -d {db_hh} {hhblits_params} -oa3m {output_hhblits}'
subprocess.run(hhblits_cmd, shell=True)
# subprocess.run(hhblits_cmd, shell=True, stdout=open(os.devnull, 'wb'))
reformat = ['reformat.pl', output_hhblits, output_reformat1]
subprocess.run(reformat)
reformat = ['reformat.pl', output_reformat1, output_reformat2]
subprocess.run(reformat)
fam_msa = "/vol/sci/bio/data/or.zuk/projects/ContactMaps/src/Periscope/data/families/xcl1_family/msa.fasta"
os.rename(output_reformat2, fam_msa)
def main():
args = parse_args()
dataset = args.dataset
LOGGER.info(f'Working on {dataset}')
path_to_upload = os.path.join(PATHS.models, 'modeller', dataset)
check_path(path_to_upload)
save_modeller_scores(dataset)
upload_folder(path_to_upload, path_to_upload.split('Periscope/')[-1])
def get_target_path(target, family=None):
if family is not None:
fam_path = os.path.join(PATHS.periscope, 'data', 'families', family,
target)
check_path(fam_path)
return fam_path
f_name = target
t_path = os.path.join(PATHS.proteins, target[1:3], f_name)
check_path(t_path)
return t_path
def main():
family = "xcl1_family"
dc = DataCreator("A0A2K6CRQ6", family=family)
msa = dc._parse_msa()
property_path = os.path.join(PATHS.periscope, "property")
check_path(property_path)
for t in msa:
try:
_ = DataCreator(t, family=family, train=False).raptor_properties
pth = _get_property_path(t, family)
dst = os.path.join(property_path, t)
shutil.copytree(pth, dst)
upload_folder(dst, dst.split('Periscope/')[-1])
except Exception:
pass
def _save_plot_matrices(model: ContactMapEstimator, predictions, family=None):
for target in predictions['logits']:
data = {}
ds = get_target_dataset(target) if family is None else family
if ds is None:
LOGGER.info(f'Problem with {target}')
continue
data_path = os.path.join(model.path, 'predictions', ds)
check_path(data_path)
target_path = os.path.join(data_path, target)
check_path(target_path)
dc = DataCreator(target, family=family)
refs_contacts = dc.refs_contacts
# pd.DataFrame(refs_contacts).to_csv(os.path.join(target_path, 'refs_contacts.csv'))
prediction = np.squeeze(predictions['logits'][target])
data['prediction'] = prediction
weights = np.squeeze(predictions['weights'][target])
data['weights'] = weights
# pd.DataFrame(prediction).to_csv(os.path.join(target_path, 'prediction.csv'))
try:
gt = dc.protein.cm
except Exception:
gt = None
data['gt'] = gt
# pd.DataFrame(gt).to_csv(os.path.join(target_path, 'gt.csv'))
if family is None:
data['alignment'] = dc.templates_aln
# data['evfold'] = dc.evfold
data['ccmpred'] = dc.ccmpred
data['templates'] = dc.k_reference_dm_test
data['seqs'] = dc.seq_refs_ss_acc
data['beff'] = dc.beff
data['refs_contacts'] = refs_contacts
pkl_save(os.path.join(target_path, 'data.pkl'), data)
upload_folder(target_path, target_path.split('Periscope/')[-1])