def load_palantir_data(smoothed=False):
fn = '../../data/external/Palantir/human_cd34_bm_rep1.h5ad'
an = anndata.read_h5ad(fn)
genes = an.var_names
cells = an.obs_names
if not smoothed:
counts = singlet.CountsTable(
data=an.raw.X.todense().T,
index=genes,
columns=cells,
)
else:
counts = singlet.CountsTable(
data=an.obsm['MAGIC_imputed_data'].T,
index=genes,
columns=cells,
)
ss = singlet.SampleSheet(an.obs)
ss['tsne_1'] = an.obsm['tsne'][:, 0]
ss['tsne_2'] = an.obsm['tsne'][:, 1]
ss['clusters'] = ss['clusters'].astype(str)
ds = singlet.Dataset(
counts_table=counts,
samplesheet=ss,
)
return ds
def load_palantir_data(smoothed=False):
fn = '../../data/external/Palantir/human_cd34_bm_rep1.h5ad'
an = anndata.read_h5ad(fn)
genes = an.var_names
cells = an.obs_names
if not smoothed:
counts = singlet.CountsTable(
data=an.raw.X.todense().T,
index=genes,
columns=cells,
)
else:
counts = singlet.CountsTable(
data=an.obsm['MAGIC_imputed_data'].T,
index=genes,
columns=cells,
)
ss = singlet.SampleSheet(an.obs)
ss['tsne_1'] = an.obsm['tsne'][:, 0]
ss['tsne_2'] = an.obsm['tsne'][:, 1]
ss['clusters'] = ss['clusters'].astype(str)
ds = singlet.Dataset(
counts_table=counts,
samplesheet=ss,
)
ds.samplesheet['Cell Subtype'] = ds.samplesheet['clusters'].replace({
'0':
'HSC',
'1':
'HSC',
'2':
'Ery-precursor',
'3':
'Mono',
'4':
'Mono-precursor',
'5':
'CLP',
'6':
'Mono',
'7':
'pDC',
'8':
'Ery',
'9':
'Mega',
})
return ds
def load_our_data():
ds = singlet.Dataset(dataset={
'path': '../../data/sequencing/me1/with_gene_names.loom',
'index_samples': 'CellID',
'index_features': 'GeneName'
}, )
ds.samplesheet['coverage'] = ds.counts.sum(axis=0)
ds.samplesheet['n_genes'] = (ds.counts >= 1).sum(axis=0)
ds.featuresheet['exp_avg'] = ds.counts.mean(axis=1)
return ds
sys.path.append('/home/fabio/university/postdoc/singlet')
os.environ['SINGLET_CONFIG_FILENAME'] = 'singlet.yml'
import singlet
if __name__ == '__main__':
fdn = '../../data/sequencing/me1/'
fn_dataset = fdn+'raw.loom'
ds = singlet.Dataset(
dataset={
'path': fn_dataset,
'index_samples': 'CellID',
'index_features': 'EnsemblID',
},
)
conv = pd.read_csv(
'../../data/gene_ensemblId_name.tsv',
sep='\t',
index_col=0,
squeeze=True,
)
print('Restrict to features with a gene name')
gids = ds.featurenames
gids = gids[gids.isin(conv.index)]
ds.query_features_by_name(gids, inplace=True)
datasetd = {
'cd137': 'anti-CD137_7dpi',
'isotype_control': 'isotype_control',
'uninfected': 'M_GV-Na_ve-Na_ve',
}
pa = argparse.ArgumentParser()
pa.add_argument('--sample', required=True, choices=datasetnames)
pa.add_argument('--normalized', action='store_true')
args = pa.parse_args()
dn = args.sample
print('Load {:} data from loom file'.format(dn))
sn = datasetd[dn]
if args.normalized:
ds = singlet.Dataset(dataset=dn + '_cpm')
# FIXME
ds.counts = singlet.CountsTable(ds.counts)
ds.counts._normalized = 'counts_per_million'
else:
ds = singlet.Dataset(dataset=dn)
ds.samplesheet['coverage'] = ds.counts.sum(axis=0)
print('Normalize cpm')
normg = ds.samplesheet['coverage'] / 1000000
ds.counts = singlet.CountsTable(ds.counts / normg)
ds.counts._normalized = 'counts_per_million'
print('Save to normalized loom file')
ds.to_dataset_file(
enh['LMO2'] = enh['Ebox_motifs']
enh = enh[tfs]
enh.columns.name = 'Motif'
return enh
if __name__ == '__main__':
fn_ds = '../../data/sequencing/me1/normalized_7tfs.loom'
if not os.path.isfile(fn_ds):
print('Read loom raw file')
fdn = '../../data/sequencing/me1/'
fn_dataset = fdn + 'raw.loom'
ds = singlet.Dataset(dataset={
'path': fn_dataset,
'index_samples': 'CellID',
'index_features': 'EnsemblID',
}, )
ds.samplesheet['coverage'] = ds.counts.sum(axis=0)
ds.samplesheet['n_genes'] = (ds.counts >= 1).sum(axis=0)
ds.featuresheet['exp_avg'] = ds.counts.mean(axis=1)
ds.counts.normalize('counts_per_ten_thousand', inplace=True)
tmp = ds.featuresheet.loc[ds.featuresheet['GeneName'].isin(tfs),
'GeneName']
dic = {val: key for key, val in tmp.items()}
idx = [dic[val] for val in tfs]
dst = ds.query_features_by_name(idx)
dst.reindex('features', 'GeneName', inplace=True)
sys.path.append('/home/fabio/university/postdoc/singlet')
import singlet
if __name__ == '__main__':
fig_fdn = '../../figures'
pa = argparse.ArgumentParser()
pa.add_argument('--save', action='store_true')
args = pa.parse_args()
print('Load data')
data_fdn = '../../data/sequencing'
fn_normalised = f'{data_fdn}/normalised.h5ad'
ds = singlet.Dataset(dataset={
'path': fn_normalised,
})
print('Loaded')
print('Load umap')
fn_umap = f'{data_fdn}/umap.tsv'
vs = pd.read_csv(fn_umap, sep='\t', index_col=0)
vs = vs.loc[ds.samplenames]
ds.obs['umap1'] = vs['umap1']
ds.obs['umap2'] = vs['umap2']
ds.obs['leiden'] = vs['leiden'].astype(str)
print('Sort by average Runx1 expression')
dsa = ds.average('samples', by='leiden')
ds.obs['cluster_new'] = ds.obs['leiden'].map({
'3': '0',
print('Export to loom file')
col_attrs = {col: metaf[col].values for col in metaf.columns}
col_attrs['CellID'] = metaf.index.values
row_attrs = {'GeneName': counts_gn.index.values}
loompy.create(
fn_loom,
counts_gn.values,
col_attrs=col_attrs,
row_attrs=row_attrs,
)
print('Load back loom file to check umap')
ds = singlet.Dataset(
dataset={
'path': fn_loom,
'index_samples': 'CellID',
'index_features': 'GeneName',
})
features = ds.feature_selection.overdispersed_within_groups('sample')
dsf = ds.query_features_by_name(features)
dsc = dsf.dimensionality.pca(n_dims=30, return_dataset='samples')
vs = dsc.dimensionality.umap()
cus = ds.samplesheet['cluster'].unique()
cmap = dict(zip(cus, sns.color_palette('husl', n_colors=len(cus))))
fig, ax = plt.subplots(figsize=(6, 4))
for cu in cus:
x, y = vs.loc[ds.samplesheet['cluster'] == cu].values.T
ax.scatter(x, y, s=30, color=cmap[cu], alpha=0.6, label=cu)
ax.legend(
loc='upper left',
genes = an.var_names
cells = an.obs_names
counts = singlet.CountsTable(
data=an.X.T,
index=genes,
columns=cells,
)
ss = singlet.SampleSheet(an.obs)
ss['tsne_1'] = an.obsm['tsne'][:, 0]
ss['tsne_2'] = an.obsm['tsne'][:, 1]
ss['clusters'] = ss['clusters'].astype(str)
ds = singlet.Dataset(
counts_table=counts,
samplesheet=ss,
)
print('Get MAGIC smoothed data')
counts = singlet.CountsTable(
data=an.obsm['MAGIC_imputed_data'].T,
index=genes,
columns=cells,
)
dsM = singlet.Dataset(
counts_table=counts,
samplesheet=ss,
)
print('Plot t-SNEs from their metadata')
genes = [
