Tutorial 2: scTenifoldNet¶

This notebook compares two expression matrices with the one-call compare_networks() API and then shows the equivalent step-wise class workflow.

In [1]:

Copied!

from scTenifold import compare_networks, scTenifoldNet
from scTenifold.data import get_test_df

x = get_test_df(n_cells=80, n_genes=40, random_state=0)
y = get_test_df(n_cells=80, n_genes=40, random_state=1)
from scTenifold import compare_networks, scTenifoldNet
from scTenifold.data import get_test_df

x = get_test_df(n_cells=80, n_genes=40, random_state=0)
y = get_test_df(n_cells=80, n_genes=40, random_state=1)

One-call workflow¶

For quick analyses, call compare_networks(). The small network settings below keep the tutorial fast; increase `n_nets, n_samp_cells, tensor rank, and tensor iterations for real analyses.

In [2]:

Copied!





result = compare_networks(
    x,
    y,
    x_label="control",
    y_label="condition",
    qc_kws={"min_lib_size": 1, "plot": False, "max_mito_ratio": 1.0, "min_percent": 0},
    network_kws={"n_nets": 2, "n_samp_cells": 25, "q": 0},
    backend="serial",
    td_kws={"K": 2, "max_iter": 20, "init": "random"},
    ma_kws={"d": 2},
)

result.head()
result = compare_networks(
    x,
    y,
    x_label="control",
    y_label="condition",
    qc_kws={"min_lib_size": 1, "plot": False, "max_mito_ratio": 1.0, "min_percent": 0},
    network_kws={"n_nets": 2, "n_samp_cells": 25, "q": 0},
    backend="serial",
    td_kws={"K": 2, "max_iter": 20, "init": "random"},
    ma_kws={"d": 2},
)

result.head()

Removed 0 cells with lib size < 1
Removed 0 outlier cells from original data
Found mitochondrial genes: ['MT-1', 'MT-10', 'MT-2', 'MT-3', 'MT-4', 'MT-5', 'MT-6', 'MT-7', 'MT-8', 'MT-9']
Removed 0 samples from original data (mt genes ratio > 1.0)
Removed 0 genes expressed in less than 0 of data
Removed 0 genes with expression values: average < 0 or sum < 0
finish QC: control
Removed 0 cells with lib size < 1
Removed 0 outlier cells from original data
Found mitochondrial genes: ['MT-1', 'MT-10', 'MT-2', 'MT-3', 'MT-4', 'MT-5', 'MT-6', 'MT-7', 'MT-8', 'MT-9']
Removed 0 samples from original data (mt genes ratio > 1.0)
Removed 0 genes expressed in less than 0 of data
Removed 0 genes with expression values: average < 0 or sum < 0
finish QC: condition
process qc finished in 0.009546581999529735 secs.
make_networks  processing time:  0.11581167100030143

make_networks  processing time:  0.11569302100087953
process nc finished in 0.23233089199948154 secs.
Using tensorly
(40, 40, 2)
tensor_decomp  processing time:  0.0044343650006339885
Using tensorly
(40, 40, 2)
tensor_decomp  processing time:  0.0041170049998981995
process td finished in 0.009537382000416983 secs.
manifold_alignment  processing time:  0.0026729829996838816
process ma finished in 0.002711033000196039 secs.
d_regulation  processing time:  0.011582894000639499
process dr finished in 0.011625193999861949 secs.

Out[2]:

	Gene	Distance	boxcox-transformed distance	Z	FC	p-value	adjusted p-value
34	NG-4	0.029244	-1.382743	2.033638	3.698975	0.054446	0.66952
31	NG-29	0.027043	-1.390411	1.787239	3.163037	0.075323	0.66952
10	NG-1	0.024698	-1.398821	1.516972	2.638354	0.104312	0.66952
37	NG-7	0.023560	-1.403003	1.382574	2.400828	0.121271	0.66952
21	NG-2	0.021716	-1.409936	1.159797	2.039616	0.153248	0.66952

Step-wise workflow¶

Use scTenifoldNet directly when you want intermediate outputs or when you want to resume from a saved run.

In [3]:

Copied!





model = scTenifoldNet(
    x,
    y,
    x_label="control",
    y_label="condition",
    qc_kws={"min_lib_size": 1, "plot": False, "max_mito_ratio": 1.0, "min_percent": 0},
    nc_kws={"n_nets": 2, "n_samp_cells": 25, "q": 0, "backend": "serial"},
    td_kws={"K": 2, "max_iter": 20, "init": "random"},
    ma_kws={"d": 2},
)

for step in ["qc", "nc", "td", "ma", "dr"]:
    model.run_step(step)

model.d_regulation.head()
model = scTenifoldNet(
    x,
    y,
    x_label="control",
    y_label="condition",
    qc_kws={"min_lib_size": 1, "plot": False, "max_mito_ratio": 1.0, "min_percent": 0},
    nc_kws={"n_nets": 2, "n_samp_cells": 25, "q": 0, "backend": "serial"},
    td_kws={"K": 2, "max_iter": 20, "init": "random"},
    ma_kws={"d": 2},
)

for step in ["qc", "nc", "td", "ma", "dr"]:
    model.run_step(step)

model.d_regulation.head()

Removed 0 cells with lib size < 1
Removed 0 outlier cells from original data
Found mitochondrial genes: ['MT-1', 'MT-10', 'MT-2', 'MT-3', 'MT-4', 'MT-5', 'MT-6', 'MT-7', 'MT-8', 'MT-9']
Removed 0 samples from original data (mt genes ratio > 1.0)
Removed 0 genes expressed in less than 0 of data
Removed 0 genes with expression values: average < 0 or sum < 0
finish QC: control
Removed 0 cells with lib size < 1
Removed 0 outlier cells from original data
Found mitochondrial genes: ['MT-1', 'MT-10', 'MT-2', 'MT-3', 'MT-4', 'MT-5', 'MT-6', 'MT-7', 'MT-8', 'MT-9']
Removed 0 samples from original data (mt genes ratio > 1.0)
Removed 0 genes expressed in less than 0 of data
Removed 0 genes with expression values: average < 0 or sum < 0
finish QC: condition
process qc finished in 0.008201310000004014 secs.
make_networks  processing time:  0.11497028999929171

make_networks  processing time:  0.1151920299998892
process nc finished in 0.2309635209994667 secs.
Using tensorly
(40, 40, 2)
tensor_decomp  processing time:  0.0044216360001883
Using tensorly
(40, 40, 2)
tensor_decomp  processing time:  0.004125004999878001
process td finished in 0.009382821999679436 secs.
manifold_alignment  processing time:  0.00213844199970481
process ma finished in 0.0021745820004070993 secs.
d_regulation  processing time:  0.010781072999634489
process dr finished in 0.010820974000125716 secs.

Out[3]:

	Gene	Distance	boxcox-transformed distance	Z	FC	p-value	adjusted p-value
34	NG-4	0.029244	-1.382743	2.033638	3.698975	0.054446	0.66952
31	NG-29	0.027043	-1.390411	1.787239	3.163037	0.075323	0.66952
10	NG-1	0.024698	-1.398821	1.516972	2.638354	0.104312	0.66952
37	NG-7	0.023560	-1.403003	1.382574	2.400828	0.121271	0.66952
21	NG-2	0.021716	-1.409936	1.159797	2.039616	0.153248	0.66952

In [4]:

Copied!





{
    "qc_shape": model.QC_dict["control"].shape,
    "n_networks": len(model.network_dict["control"]),
    "tensor_shape": model.tensor_dict["control"].shape,
    "manifold_shape": model.manifold.shape,
}
{
    "qc_shape": model.QC_dict["control"].shape,
    "n_networks": len(model.network_dict["control"]),
    "tensor_shape": model.tensor_dict["control"].shape,
    "manifold_shape": model.manifold.shape,
}

Out[4]:

{'qc_shape': (40, 80),
 'n_networks': 2,
 'tensor_shape': (40, 40),
 'manifold_shape': (80, 2)}