def read_10X(data_path, var_names='gene_symbols'):
adata = read_mtx(data_path + '/matrix.mtx').T
genes = pd.read_csv(data_path + '/genes.tsv', header=None, sep='\t')
adata.var['gene_ids'] = genes[0].values
adata.var['gene_symbols'] = genes[1].values
assert var_names == 'gene_symbols' or var_names == 'gene_ids', \
'var_names must be "gene_symbols" or "gene_ids"'
if var_names == 'gene_symbols':
var_names = genes[1]
else:
var_names = genes[0]
if not var_names.is_unique:
var_names = make_index_unique(pd.Index(var_names))
print('var_names are not unique, "make_index_unique" has applied')
adata.var_names = var_names
cells = pd.read_csv(data_path + '/barcodes.tsv', header=None, sep='\t')
adata.obs['barcode'] = cells[0].values
adata.obs_names = cells[0]
return adata
def test_make_index_unique():
index = pd.Index(["val", "val", "val-1", "val-1"])
with pytest.warns(UserWarning):
result = make_index_unique(index)
expected = pd.Index(["val", "val-2", "val-1", "val-1-1"])
assert list(expected) == list(result)
assert result.is_unique
def adata_processing_TF_link(adata,
nt_layers,
TF_list,
gene_filter_rate=0.1,
cell_filter_UMI=10000):
"""preprocess adata and get ready for TF-target gene analysis"""
n_obs, n_var = adata.n_obs, adata.n_vars
# filter genes
print(f"Original gene number: {n_var}")
gene_filter_new = (adata.layers[nt_layers[0]] >
0).sum(0) > (gene_filter_rate * n_obs)
gene_filter_tot = (adata.layers[nt_layers[1]] >
0).sum(0) > (gene_filter_rate * n_obs)
if issparse(adata.layers[nt_layers[0]]):
gene_filter_new = gene_filter_new.A1
if issparse(adata.layers[nt_layers[1]]):
gene_filter_tot = gene_filter_tot.A1
adata = adata[:, gene_filter_new * gene_filter_tot]
print(
f"Gene number after filtering: {sum(gene_filter_new * gene_filter_tot)}"
)
# filter cells
print(f"Original cell number: {n_obs}")
cell_filter = adata.layers[nt_layers[1]].sum(1) > cell_filter_UMI
if issparse(adata.layers[nt_layers[1]]):
cell_filter = cell_filter.A1
adata = adata[cell_filter, :]
print(f"Cell number after filtering: {adata.n_obs}")
# generate the expression matrix for downstream analysis
new = adata.layers[nt_layers[0]]
total = adata.layers[nt_layers[1]]
# recalculate size factor
from ..preprocessing import szFactor
adata = szFactor(adata,
method="mean-geometric-mean-total",
round_exprs=True,
total_layers=["total"])
szfactors = adata.obs["Size_Factor"][:, None]
# normalize data (size factor correction, log transform and the scaling)
if issparse(new):
new = new.A
if issparse(total):
total = total.A
new_mat = normalize_data(new, szfactors, pseudo_expr=0.1)
tot_mat = normalize_data(total, szfactors, pseudo_expr=0.1)
new_mat = pd.DataFrame(new_mat,
index=adata.obs_names,
columns=adata.var_names)
tot_mat = pd.DataFrame(tot_mat,
index=adata.obs_names,
columns=adata.var_names)
# compute the labeling reads rate in each cell
obs = adata.obs
var = adata.var
var.loc[:, "gene_short_name"] = make_index_unique(
var.loc[:, "gene_short_name"].astype("str"))
obs.loc[:, "labeling_rate"] = adata.layers["new"].sum(
1) / adata.layers["total"].sum(1)
# extract the TF matrix
var_TF = var.query("gene_short_name in @TF_list")
print(f"\nNumber of TFs found in the list: {var_TF.shape[0]}")
TF_matrix = tot_mat.loc[:, var_TF.loc[:, "gene_id"]]
TF_matrix.columns = var_TF.loc[:, "gene_short_name"]
return (tot_mat.T, new_mat.T, obs, var, var_TF, TF_matrix.T)