def alignment_score_kernel(true_states: str,
pred_states: str,
kernel_widths: list,
query_offset: int = 0,
hit_offset: int = 0,
no_gaps=True):
""" Computes ROC statistics on alignment.
Parameters
----------
true_states : str
Ground truth state string
pred_states : str
Predicted state string
"""
pred_states = list(map(tmstate_f, pred_states))
true_states = list(map(tmstate_f, true_states))
pred_edges = states2edges(pred_states)
true_edges = states2edges(true_states)
# add offset to account for local alignments
true_edges = list(map(tuple, np.array(true_edges)))
pred_edges = np.array(pred_edges)
pred_edges[:, 0] += query_offset
pred_edges[:, 1] += hit_offset
pred_edges = list(map(tuple, pred_edges))
if no_gaps:
pred_edges = filter_gaps(pred_states, pred_edges)
true_edges = filter_gaps(true_states, true_edges)
res = []
for k in kernel_widths:
r = roc_edges_kernel_identity(true_edges, pred_edges, k)
res.append(r)
return res
def alignment_score(true_states: str, pred_states: str):
"""
Computes ROC statistics on alignment
Parameters
----------
true_states : str
Ground truth state string
pred_states : str
Predicted state string
"""
pred_states = list(map(tmstate_f, pred_states))
true_states = list(map(tmstate_f, true_states))
pred_edges = states2edges(pred_states)
true_edges = states2edges(true_states)
stats = roc_edges(true_edges, pred_edges)
return stats
def validation_stats(self, x, y, xlen, ylen, gen, states, A, predA, theta,
gap, batch_idx):
statistics = []
for b in range(len(xlen)):
# TODO: Issue #47
x_str = decode(
list(x[b, :xlen[b]].squeeze().cpu().detach().numpy()),
self.tokenizer.alphabet)
y_str = decode(
list(y[b, :ylen[b]].squeeze().cpu().detach().numpy()),
self.tokenizer.alphabet)
decoded, _ = next(gen)
pred_x, pred_y, pred_states = list(zip(*decoded))
pred_states = np.array(list(pred_states))
true_states = states[b].cpu().detach().numpy()
pred_edges = states2edges(pred_states)
true_edges = states2edges(true_states)
pred_edges = filter_gaps(pred_states, pred_edges)
true_edges = filter_gaps(true_states, true_edges)
stats = roc_edges(true_edges, pred_edges)
if random.random() < self.hparams.visualization_fraction:
Av = A[b].cpu().detach().numpy().squeeze()
pv = predA[b].cpu().detach().numpy().squeeze()
tv = theta[b].cpu().detach().numpy().squeeze()
gv = gap[b].cpu().detach().numpy().squeeze()
fig, _ = alignment_visualization(Av, pv, tv, gv, xlen[b],
ylen[b])
self.logger.experiment.add_figure(
f'alignment-matrix/{batch_idx}/{b}',
fig,
self.global_step,
close=True)
try:
text = alignment_text(x_str, y_str, pred_states,
true_states, stats)
self.logger.experiment.add_text(
f'alignment/{batch_idx}/{b}', text, self.global_step)
except Exception as e:
print(predA[b])
print(A[b])
print(theta[b])
print(xlen[b], ylen[b])
raise e
statistics.append(stats)
return statistics
def __getitem__(self, i):
""" Gets alignment pair.
Parameters
----------
i : int
Index of item
Returns
-------
gene : torch.Tensor
Encoded representation of protein of interest
pos : torch.Tensor
Encoded representation of protein that aligns with `gene`.
states : torch.Tensor
Alignment string
alignment_matrix : torch.Tensor
Ground truth alignment matrix
path_matrix : torch.Tensor
Pairwise path distances, where the smallest distance
to the path is computed for every element in the matrix.
"""
gene = self.pairs.iloc[i]['chain1']
pos = self.pairs.iloc[i]['chain2']
st = self.pairs.iloc[i]['alignment']
states = list(map(tmstate_f, st))
if self.clip_ends:
gene, pos, states, st = clip_boundaries(gene, pos, states, st)
if self.pad_ends:
states = [m] + states + [m]
states = torch.Tensor(states).long()
gene = self.tokenizer(str.encode(gene))
pos = self.tokenizer(str.encode(pos))
gene = torch.Tensor(gene).long()
pos = torch.Tensor(pos).long()
alignment_matrix = torch.from_numpy(states2matrix(states))
path_matrix = torch.empty(*alignment_matrix.shape)
g_mask = torch.ones(*alignment_matrix.shape)
if self.construct_paths:
pi = states2edges(states)
path_matrix = torch.from_numpy(path_distance_matrix(pi))
path_matrix = reshape(path_matrix, len(gene), len(pos))
if self.mask_gaps:
g_mask = torch.from_numpy(gap_mask(st)).bool()
alignment_matrix = reshape(alignment_matrix, len(gene), len(pos))
g_mask = reshape(g_mask, len(gene), len(pos))
if not self.return_names:
return gene, pos, states, alignment_matrix, path_matrix, g_mask
else:
gene_name = self.pairs.iloc[i]['chain1_name']
pos_name = self.pairs.iloc[i]['chain2_name']
return (gene, pos, states, alignment_matrix, path_matrix, g_mask,
gene_name, pos_name)
def alignment_score(true_states: str, pred_states: str, no_gaps=True):
""" Computes ROC statistics on alignment.
Parameters
----------
true_states : str
Ground truth state string
pred_states : str
Predicted state string
no_gaps : bool
Removes all gaps before computing ROC stats.
"""
pred_states = list(map(tmstate_f, pred_states))
true_states = list(map(tmstate_f, true_states))
pred_edges = states2edges(pred_states)
true_edges = states2edges(true_states)
if no_gaps:
pred_edges = filter_gaps(pred_states, pred_edges)
true_edges = filter_gaps(true_states, true_edges)
stats = roc_edges(true_edges, pred_edges)
return stats