def entropy_batch_mixing(self, name, verbose=False, **kwargs):
if self.gene_dataset.n_batches == 2:
latent, batch_indices, labels = get_latent(self.model,
self.data_loaders[name])
be_score = entropy_batch_mixing(latent, batch_indices, **kwargs)
if verbose:
print("Entropy batch mixing :", be_score)
return be_score
def harmonization_stat(model, data_loader, keys, pop1, pop2):
latent, batch_indices, labels = get_latent(model, data_loader)
batch_indices = np.concatenate(batch_indices)
sample = sample_by_batch(batch_indices, 2000)
sample_2batch = sample[(batch_indices[sample] == pop1) +
(batch_indices[sample] == pop2)]
batch_entropy = entropy_batch_mixing(latent[sample_2batch, :],
batch_indices[sample_2batch])
print("Entropy batch mixing : %f.3" % batch_entropy)
sample = sample_by_batch(labels, 200)
res = knn_purity_avg(latent[sample, :],
labels.astype('int')[sample],
keys,
acc=True)
print("Average knn purity : %f.3" % np.mean([x[1] for x in res]))
return (batch_entropy, res)
print("dataset 2 has %d cells" % (np.sum(count[1])))
print(
"correlation between the cell-type composition of the subsampled dataset is %.3f"
% correlation)
sub_dataset = deepcopy(gene_dataset)
sub_dataset.update_cells(np.concatenate(cells))
vae = VAE(sub_dataset.nb_genes,
n_batch=sub_dataset.n_batches,
n_labels=sub_dataset.n_labels,
n_hidden=128,
dispersion='gene')
infer = VariationalInference(vae, sub_dataset, use_cuda=use_cuda)
infer.train(n_epochs=250)
latent, batch_indices, labels = infer.get_latent('sequential')
keys = sub_dataset.cell_types
batch_entropy = entropy_batch_mixing(latent, batch_indices)
print("Entropy batch mixing :", batch_entropy)
sample = select_indices_evenly(1000, labels)
res = knn_purity_avg(latent[sample, :],
labels[sample].astype('int'),
keys=keys,
acc=True)
print('average classification accuracy per cluster')
for x in res:
print(x)
knn_acc = np.mean([x[1] for x in res])
print("average KNN accuracy:", knn_acc)
res = clustering_scores(
np.asarray(latent)[sample, :], labels[sample], 'knn',
len(np.unique(labels[sample])))
for x in res:
latent = np.genfromtxt('../macosko_regev.CCA.txt')
label = np.genfromtxt('../macosko_regev.CCA.label.txt',dtype='str')
keys = gene_dataset.cell_types
batch_indices = np.genfromtxt('../macosko_regev.CCA.batch.txt')
elif model_type == 'Combat':
COMBAT = COMBAT()
latent = COMBAT.combat_pca(gene_dataset)
latent = latent.T
batch_indices = np.concatenate(gene_dataset.batch_indices)
labels = np.concatenate(gene_dataset.labels)
keys = gene_dataset.cell_types
sample = select_indices_evenly(2000,batch_indices)
batch_entropy = entropy_batch_mixing(latent[sample, :], batch_indices[sample])
print("Entropy batch mixing :", batch_entropy)
sample = select_indices_evenly(1000,labels)
res = knn_purity_avg(
latent[sample, :], labels[sample],
keys=keys[np.unique(labels)], acc=True
)
print('average classification accuracy per cluster',np.mean([x[1] for x in res]))
for x in res:
print(x)
res = clustering_scores(np.asarray(latent)[sample,:],labels[sample],'knn',len(np.unique(labels[sample])))
for x in res:
def run_benchmarks(dataset_name,
model=VAE,
n_epochs=1000,
lr=1e-3,
use_batches=False,
use_cuda=True,
show_batch_mixing=True,
benchmark=False,
tt_split=0.9,
unit_test=False):
# options:
# - gene_dataset: a GeneExpressionDataset object
# call each of the 4 benchmarks:
# - log-likelihood
# - imputation
# - batch mixing
# - cluster scores
gene_dataset = load_datasets(dataset_name, unit_test=unit_test)
example_indices = np.random.permutation(len(gene_dataset))
tt_split = int(tt_split * len(gene_dataset)) # 90%/10% train/test split
data_loader_train = DataLoader(gene_dataset,
batch_size=128,
pin_memory=use_cuda,
sampler=SubsetRandomSampler(
example_indices[:tt_split]),
collate_fn=gene_dataset.collate_fn)
data_loader_test = DataLoader(gene_dataset,
batch_size=128,
pin_memory=use_cuda,
sampler=SubsetRandomSampler(
example_indices[tt_split:]),
collate_fn=gene_dataset.collate_fn)
vae = model(gene_dataset.nb_genes,
n_batch=gene_dataset.n_batches * use_batches,
n_labels=gene_dataset.n_labels,
use_cuda=use_cuda)
stats = train(vae,
data_loader_train,
data_loader_test,
n_epochs=n_epochs,
lr=lr,
benchmark=benchmark)
if isinstance(vae, VAE):
best_ll = adapt_encoder(vae,
data_loader_test,
n_path=1,
n_epochs=1,
record_freq=1)
print("Best ll was :", best_ll)
# - log-likelihood
print("Log-likelihood Train:", stats.history["LL_train"][stats.best_index])
print("Log-likelihood Test:", stats.history["LL_test"][stats.best_index])
# - imputation
imputation_test = imputation(vae, data_loader_test)
print("Imputation score on test (MAE) is:", imputation_test.item())
# - batch mixing
if gene_dataset.n_batches == 2:
latent, batch_indices, labels = get_latent(vae, data_loader_train)
print(
"Entropy batch mixing :",
entropy_batch_mixing(latent.cpu().numpy(),
batch_indices.cpu().numpy()))
if show_batch_mixing:
show_t_sne(
latent.cpu().numpy(),
np.array([batch[0] for batch in batch_indices.cpu().numpy()]))
# - differential expression
if type(gene_dataset) == CortexDataset:
get_statistics(vae, data_loader_train, M_sampling=1,
M_permutation=1) # 200 - 100000