def main(argv):
if len(argv) > 1:
raise app.UsageError('Too many command-line arguments.')
# Set random seed
np.random.seed(FLAGS.seed)
torch.manual_seed(FLAGS.seed)
if torch.cuda.is_available(): # cuda device
device = 'cuda'
torch.cuda.set_device(FLAGS.gpu)
else:
device = 'cpu'
adata, trainloader, testloader = load_dataset(
FLAGS.input_path,
batch_name='batch',
min_genes=FLAGS.min_peaks,
min_cells=FLAGS.min_cells,
batch_size=FLAGS.batch_size,
n_top_genes=FLAGS.n_feature,
log=None,
)
input_dim = adata.shape[1]
tf.io.gfile.makedirs(FLAGS.output_path)
dims = [input_dim, FLAGS.latent, FLAGS.encode_dim, FLAGS.decode_dim]
model = SCALE(dims, n_centroids=FLAGS.n_centroids)
print(model)
print('\n## Training Model ##')
model.init_gmm_params(testloader)
model.fit(trainloader,
lr=FLAGS.lr,
weight_decay=FLAGS.weight_decay,
verbose=FLAGS.verbose,
device=device,
max_iter=FLAGS.max_iter,
outdir=FLAGS.output_path)
adata.obsm['latent'] = model.encodeBatch(testloader,
device=device,
out='z')
dr = pd.DataFrame(adata.obsm['latent'], index=adata.obs_names)
with tf.io.gfile.GFile(os.path.join(FLAGS.output_path, 'SCALE.csv'),
'w') as f:
dr.to_csv(f)
print(
"\n**********************************************************************"
)
print(
" SCALE: Single-Cell ATAC-seq analysis via Latent feature Extraction")
print(
"**********************************************************************\n"
)
print("======== Parameters ========")
print(
'Cell number: {}\nInput_dim: {}\nn_centroids: {}\nEpoch: {}\nSeed: {}\nDevice: {}'
.format(cell_num, input_dim, k, epochs, args.seed, args.device))
print("============================")
dims = [input_dim, config.latent, config.encode_dim, config.decode_dim]
model = SCALE(dims, n_centroids=k, device=device)
model.to(device)
data = data.to(device)
if not args.pretrain:
print('\n## Training Model ##')
t0 = time.time()
model.init_gmm_params(data)
model.fit(dataloader,
lr=lr,
weight_decay=config.weight_decay,
epochs=epochs,
verbose=args.verbose,
print_interval=config.print_interval)
print('\nRunning Time: {:.2f} s'.format(time.time() - t0))
else:
print('\n## Loading Model {} ##\n'.format(args.pretrain))
lr = args.lr
k = args.n_centroids
outdir = args.outdir + '/'
if not os.path.exists(outdir):
os.makedirs(outdir)
print("\n======== Parameters ========")
print(
'Cell number: {}\nPeak number: {}\nn_centroids: {}\nmax_iter: {}\nbatch_size: {}\ncell filter by peaks: {}\npeak filter by cells: {}'
.format(cell_num, input_dim, k, args.max_iter, batch_size,
args.min_peaks, args.min_cells))
print("============================")
dims = [input_dim, args.latent, args.encode_dim, args.decode_dim]
model = SCALE(dims, n_centroids=k)
print(model)
if not args.pretrain:
print('\n## Training Model ##')
model.init_gmm_params(testloader)
model.fit(
trainloader,
lr=lr,
weight_decay=args.weight_decay,
verbose=args.verbose,
device=device,
max_iter=args.max_iter,
# name=name,
outdir=outdir)
torch.save(model.state_dict(), os.path.join(outdir,
"\n**********************************************************************"
)
print(
" SCALE: Single-Cell ATAC-seq Analysis via Latent feature Extraction")
print(
"**********************************************************************\n"
)
print("======== Parameters ========")
print(
'Cell number: {}\nPeak number: {}\nn_centroids: {}\nmax_iter: {}\nbatch_size: {}\ncell filter by peaks: {}\nrare peak filter: {}\ncommon peak filter: {}'
.format(cell_num, input_dim, k, args.max_iter, batch_size,
args.min_peaks, args.low, args.high))
print("============================")
dims = [input_dim, args.latent, args.encode_dim, args.decode_dim]
model = SCALE(dims, n_centroids=k)
# print(model)
if not args.pretrain:
print('\n## Training Model ##')
model.init_gmm_params(testloader)
model.fit(trainloader,
lr=lr,
weight_decay=args.weight_decay,
verbose=args.verbose,
device=device,
max_iter=args.max_iter,
name=name,
outdir=outdir)
# torch.save(model.to('cpu').state_dict(), os.path.join(outdir, 'model.pt')) # save model
else:
args = parser.parse_args()
if args.version == 2:
from scale import SCALE
adata = SCALE(args.data_list,
batch_categories=args.batch_categories,
profile=args.profile,
join=args.join,
batch_key=args.batch_key,
min_features=args.min_features,
min_cells=args.min_cells,
n_top_features=args.n_top_features,
batch_size=args.batch_size,
lr=args.lr,
max_iteration=args.max_iteration,
impute=args.impute,
batch_name=args.batch_name,
seed=args.seed,
gpu=args.gpu,
outdir=args.outdir,
projection=args.projection,
chunk_size=args.chunk_size,
ignore_umap=args.ignore_umap,
repeat=args.repeat,
verbose=True,
assess=args.assess)
elif args.version == 1:
from extensions.scale import SCALE_v1
adata = SCALE_v1(