ds = Dataset(
counts_table=counts,
featuresheet=featuresheet,
samplesheet=samplesheet,
)
print('Add normalized virus counts')
ds.samplesheet['virus_reads_per_million'] = 1e6 * ds.samplesheet[
'n_reads_virus'] / ds.samplesheet['n_reads']
ds.samplesheet['log_virus_reads_per_million'] = np.log10(
0.1 + ds.samplesheet['virus_reads_per_million'])
print('Filter low-quality cells')
n_reads_min = args.n_reads_min
ds.query_samples_by_metadata('n_reads > @n_reads_min',
local_dict=locals(),
inplace=True)
print('Limit to decently expressed genes')
ind = (ds.counts > args.n_cpm_min_genes[0]).sum(
axis=1) >= args.n_cpm_min_genes[1]
ds.counts = ds.counts.loc[ind]
print('Ignore genes with multiple IDs')
from collections import Counter
genec = Counter(ds.featuresheet['GeneName'].values)
genes_multiple = [k for k, v in genec.items() if v > 1]
ds.featuresheet = ds.featuresheet.loc[~ds.featuresheet['GeneName'].
isin(genes_multiple)]
print('Translate to gene names')
counts_table='dengue',
samplesheet='virus',
featuresheet='humanGC38',
)
ds.query_samples_by_counts('total >= 50000', inplace=True)
ds.samplesheet.rename(columns={'time [h]': 'time'}, inplace=True)
cov = ds.samplesheet['coverage'] = ds.counts.sum(axis=0)
ds.counts.normalize('counts_per_million', inplace=True)
n = ds.samplesheet['numberDengueReads'].astype(int)
ds.samplesheet['virus_reads_per_million'] = 1e6 * n / (cov + n)
ds.counts.log(inplace=True)
# Only select cells without virus
ds.query_samples_by_metadata('virus_reads_per_million < 0.1', inplace=True)
# Check table with number of cells
table = (ds.samplesheet.groupby(
['time', 'MOI']).count().iloc[:,
0].unstack().fillna(0).astype(int).loc[[
'4', '12', '24', '48'
]])
print('Selecting only early 2 time points')
# The rest has too few uninfected cells
ds.query_samples_by_metadata('time in ["4", "12"]', inplace=True)
dsm = ds.split('MOI')
ks = dsm['0'].compare(dsm['1'])['P-value']
# Get the top hits for GO analysis
date: 07/08/17
content: Test Dataset class.
'''
# Script
if __name__ == '__main__':
# NOTE: an env variable for the config file needs to be set when
# calling this script
from singlet.dataset import Dataset
ds = Dataset(
samplesheet='example_sheet_tsv',
counts_table='example_table_tsv')
print('Query samples by metadata')
ds_tmp = ds.query_samples_by_metadata(
'experiment == "test_pipeline"',
inplace=False)
assert(tuple(ds_tmp.samplenames) == ('test_pipeline',))
print('Done!')
print('Query sample by counts in one gene')
ds_tmp = ds.query_samples_by_counts('KRIT1 > 100', inplace=False)
assert(tuple(ds_tmp.samplenames) == ('third_sample',))
print('Done!')
print('Query sample by total counts')
ds_tmp = ds.query_samples_by_counts('total < 3000000', inplace=False)
assert(tuple(ds_tmp.samplenames) == ('second_sample',))
print('Done!')
print('Query sample by mapped counts')
sep='\t',
index_col=0,
dtype={0: str},
))
print('Build dataset')
ds = Dataset(
counts_table=counts,
featuresheet=featuresheet,
samplesheet=samplesheet,
)
print('Filter low-quality cells')
n_reads_min = args.n_reads_min
ds.query_samples_by_metadata('n_reads > @n_reads_min',
local_dict=locals(),
inplace=True)
if not args.keep2:
print('Filter out sample 2-uninfected (low quality)')
ds.query_samples_by_metadata('biosample != "2-uninfected"',
inplace=True)
print('Add normalized virus counts')
ds.samplesheet['virus_reads_per_million'] = 1e6 * ds.samplesheet[
'n_reads_virus'] / ds.samplesheet['n_reads']
ds.samplesheet['log_virus_reads_per_million'] = np.log10(
0.1 + ds.samplesheet['virus_reads_per_million'])
print('Limit to decently expressed genes')
ind = (ds.counts >= 10).sum(axis=1) >= 10
'$10^5$',
])
ax.set_ylabel('{:} per million reads'.format(gname))
ax.set_xlabel('')
for tk in ax.get_xticklabels():
tk.set_rotation(300)
return fig
# QC all cells
fig = plot_qcs(ds)
fig.suptitle('All cells')
plt.tight_layout(rect=(0, 0, 1, 0.96))
# Filter cells and reQC
n_reads_min = args.n_reads_min
dsgood = ds.query_samples_by_metadata('n_reads > @n_reads_min',
local_dict=locals())
fig = plot_qcs(dsgood)
fig.suptitle('Only cells with {:d}+ reads'.format(n_reads_min))
plt.tight_layout(rect=(0, 0, 1, 0.96))
## Probably not worth saving, we only trash ~2,500 out of ~8,700 cells
#if args.save:
# dsgood.counts.to_csv(
# '../../data/mouse_mCMV_1/all/dataframe_n_reads_min_{:}.tsv'.format(n_reads_min),
# index=True, sep='\t')
plt.ion()
plt.show()
sharex=True,
sharey=True,
figsize=(8, 4))
dsr.plot.scatter_reduced_samples(vs,
color_by='dbscan',
ax=axs[0],
zorder=10)
dsr.plot.scatter_reduced_samples(vs,
color_by='kmeans',
ax=axs[1],
zorder=10)
axs[0].set_title('DBSCAN')
axs[1].set_title('K-means, 7 clusters')
plt.tight_layout()
ds.samplesheet['cluster'] = dsr.samplesheet['kmeans']
ds_dict = ds.split(phenotypes=['cluster'])
genes_by_cluster = {}
for key, dsi in ds_dict.items():
dso = ds.query_samples_by_metadata('cluster!=@key',
local_dict=locals())
genes_by_cluster[key] = dsi.compare(dso)['P-value'].nsmallest(10).index
assert (genes_by_cluster[1][:3].tolist() == [
'ENSG00000138085', 'ENSG00000184076', 'ENSG00000116459'
])
plt.show()
pa.add_argument('--keep2',
action='store_true',
help='Keep sample 2-uninfected despite low quality')
args = pa.parse_args()
print('Load dataset')
ds = Dataset(
counts_table='combined',
featuresheet='combined',
samplesheet='combined',
)
print('Filter low-quality cells')
n_reads_min = args.n_reads_min
ds.query_samples_by_metadata('n_reads > @n_reads_min',
local_dict=locals(),
inplace=True)
if not args.keep2:
print('Filter out sample 2-uninfected (low quality)')
ds.query_samples_by_metadata('biosample != "2-uninfected"',
inplace=True)
print('Add normalized virus counts')
ds.samplesheet['virus_reads_per_million'] = 1e6 * ds.samplesheet[
'n_reads_virus'] / ds.samplesheet['n_reads']
ds.samplesheet['log_virus_reads_per_million'] = np.log10(
0.1 + ds.samplesheet['virus_reads_per_million'])
print('Limit to decently expressed genes')
ind = (ds.counts >= 10).sum(axis=1) >= 10
