fold1 = next(KFold(n_splits=test_n_splits, shuffle=True).split(isect))
tests = np.array([v for i, v in enumerate(isect) if i in fold1[1]])
np.save(Path(f'data/test/scop40_{n_splits}fold_sf{sf_sunid}_testdata_{test_n_splits}fold.npy'), tests)
# select domain sids only in scop40 for training data
train = np.setdiff1d(samples, tests)
np.save(Path(f'data/train/scop40_{n_splits}fold_trainingdata.npy'), train)
# {sf: [sid]} list for making alignment pairs in the same superfamily
hie = {}
for i in train:
dom = scop100_hie.getDomainBySid(i)
if dom:
sf = dom.getAscendent('sf').sccs
else:
# FIX: Why nothing?
continue
if sf in hie:
hie[sf].append(i)
else:
hie[sf] = [i]
pickle.dump(hie, Path(f'data/train/scop40_{n_splits}fold_sf{sf_sunid}_hie.pkl').open('wb'))
else:
train = np.array([x for x in scop40 if x not in test_data])
np.save(Path(f'data/train/scop40_{n_splits}fold.npy'), train)
test = np.array(test_data)
np.save(Path(f'data/test/scop40_{n_splits}fold.npy'), test)
hie = {}
for sf in scop100_hie.getRoot().getDescendents('sf'):
isect = np.intersect1d(train, np.array([x.sid for x in sf.getDescendents('px')]))
hie[sf.sunid] = isect.tolist()
pickle.dump(hie, Path(f'data/train/scop40_{n_splits}fold_hie.pkl').open('wb'))
