def correct(datasets_full, genes_list, hvg=HVG, verbose=VERBOSE,
sigma=SIGMA, ds_names=None):
datasets, genes = merge_datasets(datasets_full, genes_list)
datasets_dimred, genes = process_data(datasets, genes, hvg=hvg)
datasets_dimred = assemble(
datasets_dimred, # Assemble in low dimensional space.
expr_datasets=datasets, # Modified in place.
verbose=verbose, knn=KNN, sigma=sigma, approx=APPROX,
ds_names=ds_names
)
return datasets, genes
def panorama(datasets_full, genes_list):
datasets, genes = merge_datasets(datasets_full, genes_list)
datasets_dimred, genes = process_data(datasets, genes)
# Connected components form panoramas.
panoramas = connect(datasets_dimred,
knn=KNN,
approx=APPROX,
verbose=VERBOSE)
if VERBOSE:
print(panoramas)
return panoramas
def panorama(datasets_full, genes_list):
if VERBOSE:
print('Processing and reducing dimensionality...')
datasets, genes = merge_datasets(datasets_full, genes_list)
datasets_dimred, genes = process_data(datasets, genes)
if VERBOSE:
print('Finding panoramas...')
panoramas = connect(datasets_dimred)
if VERBOSE:
print(panoramas)
return panoramas
genes_hvg, _ = hvg([dataset.X], dataset.var['gene_symbols'],
'dispersion')
if hv_genes is None:
hv_genes = set(
[g for gene in genes_hvg[:12000] for g in gene.split(';')])
else:
hv_genes &= set(
[g for gene in genes_hvg[:12000] for g in gene.split(';')])
tprint('{}: {}'.format(all_namespaces[i], len(hv_genes)))
sys.stdout.flush()
# Keep only those highly variable genes.
Xs, genes = merge_datasets(
[dataset.X for dataset in all_datasets],
[dataset.var['gene_symbols'] for dataset in all_datasets],
keep_genes=hv_genes,
ds_names=all_namespaces,
verbose=True)
[tprint(X.shape[0]) for X in Xs]
X = vstack(Xs)
X = X.log1p()
cell_types = np.concatenate(
[dataset.obs['cell_types'] for dataset in all_datasets], axis=None)
ages = np.concatenate([dataset.obs['ages'] for dataset in all_datasets],
axis=None)
cds = [
PanDAG(
continue
u, p = mannwhitneyu(A[:, idx], B[:, idx])
p_vals.append(p)
reject, q_vals, _, _ = multipletests(p_vals, method='bonferroni')
for idx, gene in enumerate(genes):
if reject[idx]:
print('{}\t{}'.format(gene, q_vals[idx]))
if __name__ == '__main__':
datasets, genes_list, n_cells = load_names(data_names)
monocytes, mono_genes = datasets[:4], genes_list[:4]
monocytes, mono_genes = merge_datasets(monocytes, mono_genes)
datasets = [vstack(monocytes)] + datasets[4:]
genes_list = [mono_genes] + genes_list[4:]
data_names = ['data/macrophage/monocytes_seqwell'] + data_names[4:]
datasets, genes = merge_datasets(datasets, genes_list)
datasets_dimred, genes = process_data(datasets, genes)
_, A, _ = find_alignments_table(datasets_dimred)
time = np.array([0, 0, 3, 3, 6, 6]).reshape(-1, 1)
time_align_correlate(A, time)
x = np.array([0, 0, 1, 1, 2, 2]).reshape(-1, 1)
y = [-.1, .1, -.1, .1, -.1, .1]
time_align_visualize(A, x, y, namespace=NAMESPACE)
from time_align import time_align_correlate, time_align_visualize, time_dist
NAMESPACE = 'dendritic'
data_names = [
'data/dendritic/unstimulated',
'data/dendritic/unstimulated_repl',
'data/dendritic/lps_1h',
'data/dendritic/lps_2h',
'data/dendritic/lps_4h',
'data/dendritic/lps_4h_repl',
'data/dendritic/lps_6h',
]
if __name__ == '__main__':
datasets, genes_list, n_cells = load_names(data_names)
datasets, genes = merge_datasets(datasets, genes_list)
datasets_dimred, genes = process_data(datasets, genes)
_, A, _ = find_alignments_table(datasets_dimred)
time = np.array([0, 0, 1, 2, 4, 4, 6]).reshape(-1, 1)
time_align_correlate(A, time)
time_dist(datasets_dimred, time)
x = np.array([0, 0, 1, 2, 3, 3, 4]).reshape(-1, 1)
y = [-.1, .1, 0, 0, -.1, .1, 0]
time_align_visualize(A, x, y, namespace=NAMESPACE)
else:
raise ValueError('Unhandled age {}'.format(age_str))
ages += [age] * datasets[i].shape[0]
inj = 'demyelination' if 'Lysolecithin' in meta['treatment'] else 'none'
injured += [inj] * datasets[i].shape[0]
cell_types += ['{}_{}'.format(age_str, inj)] * datasets[i].shape[0]
return np.array(cell_types), np.array(ages), np.array(injured)
datasets, genes_list, n_cells = load_names(data_names, norm=False)
cell_types, ages, injured = load_meta(datasets)
datasets, genes = merge_datasets(datasets,
genes_list,
union=True,
verbose=False)
X = vstack(datasets)
qc_idx = [i for i, s in enumerate(np.sum(X != 0, axis=1)) if s >= 500]
tprint('Found {} valid cells among all datasets'.format(len(qc_idx)))
X = X[qc_idx]
cell_types = cell_types[qc_idx]
ages = ages[qc_idx]
injured = injured[qc_idx]
if not os.path.isfile('data/dimred/{}_{}.txt'.format(DR_METHOD, NAMESPACE)):
mkdir_p('data/dimred')
tprint('Dimension reduction with {}...'.format(DR_METHOD))
X_dimred = reduce_dimensionality(normalize(X), dim_red_k=DIMRED)
'data/pbmc/10x/b_cells',
'data/pbmc/10x/cd14_monocytes',
'data/pbmc/10x/cd34',
'data/pbmc/10x/cd4_t_helper',
'data/pbmc/10x/cd56_nk',
'data/pbmc/10x/cytotoxic_t',
'data/pbmc/10x/memory_t',
'data/pbmc/10x/naive_cytotoxic_t',
'data/pbmc/10x/naive_t',
'data/pbmc/10x/regulatory_t',
'data/pbmc/68k',
]
datasets, genes_list, n_cells = load_names(data_names)
Xs, genes = merge_datasets(datasets[:], genes_list, ds_names=data_names)
uniq_cell_types = [
'CD19+_B',
'CD14+_Monocyte',
'CD34+',
'CD4+_T_Helper2',
'CD56+_NK',
'CD8+_Cytotoxic_T',
'CD4+/CD45RO+_Memory',
'CD8+/CD45RA+_Naive_Cytotoxic',
'CD4+/CD45RA+/CD25-_Naive_T',
'CD4+/CD25_T_Reg',
]
cell_types = []
