P0PCorNS (Perturbation to 0 to Predict Correlated Network Stability) performs informative gene knockouts (One gene was in silico knocked out each time ), and uses BNW (Ziebarth et al. Bioinformatics, 2013.) to generate the Bayesian Networks, then calculates the Jensen-Shannon divergence (JSD) between networks. Finally, P0PCorNS can provide a order list of gene impacts.
A gene that exhibits a higher informative impact would play an more importent role in in the gene regulatory networks (potentially function more upstream in a linear regulatory order or be the hub gene) , and will be ranked higher for gene manipulation verification experiments.
P0PCorNS has been tested in Red Hat Enterprise Linux release 8.8 and is based on Python 3. To install P0PCorNS, just download this repository.
To test and get familiar with P0PCorNS, you can copy the demo files we provide under the 'example/input_demo' directory to a folder(for example, the "input" folder), then run the following codes:
python P0PCorNS.py <csv_file_path> [k]
You can also compare your output to the results in the directory 'example/output_demo'.
- demo_short.csv
demo_short.csv contains the foldchange information of 4 genes.
This file is just for quick test. The running time of 4 genes and 3 stages in both Gene&Stage mode and Gene_only mode is less than 1 min in the testing server.
st01,st12,st23,rel,fmf,pp1,pna
1,1,1,1,1,1,1
1,2,2,1.95934,12.2566,0.970340666,1.80615
1,2,3,1.976248755,11.80867992,1.088135298,1.256936979
1,2,3,2.218305097,10.73544157,1.17096892,2.582973777
The first row are variable names.
The second row describes the type of each variable: 1 is for each continuous variable.
The remaining rows are the foldchange information of the variable (Stage specific expression data collected from real experiments).
Note: P0PCorNS also provide a demo_long.csv as an example of more complicated cases.
The demo_long.csv contains the foldchange information of 11 genes.
The running time of 11 genes and 5 stages in both Gene&Stage mode and Gene_only mode is about 40 min in the testing server.
st01,st12,st23,st34,st45,rel,fmf,pp1,pna,c837,adv,asm,vad,MT1,MT2,bk1
1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1
1,2,2,2,2,1.95934,12.2566,0.970340666,1.80615,3.48704,0.864781922,1.10295,1.39447,1.750521515,1.98014,0.32091
1,2,3,3,3,1.976248755,11.80867992,1.088135298,1.256936979,10.50862851,1.101398661,1.302777735,1.515060619,1.190752282,1.526687561,0.559462795
1,2,3,4,4,2.218305097,10.73544157,1.17096892,2.582973777,10.50200323,1.036128621,1.662780757,2.321259143,0.020534607,1.393654735,0.806710002
1,2,3,4,5,2.231886212,9.65814396,0.92137892,2.984532145,10.73712334,-0.115881132,1.589796151,2.67546936,-0.149555393,1.978355461,-0.494486519
2,3,4,5,5,2.98139685,12.44567093,1.11011892,3.282725138,10.95434666,-1.650521132,1.879659864,2.480042579,1.524784607,2.594232992,-0.728818972
- Gene&Stage mode
Gene&Stage mode will use the stage information and foldchange information of genes to generate the bayesian network.
- Gene_only mode
Gene_only mode will only use the foldchange information of genes to generate the bayesian network.
That is, for the Gene_only mode, P0PCorNS will only use the gene columns as the input, like the following:
rel,fmf,pp1,pna
1,1,1,1
1.95934,12.2566,0.970340666,1.80615
1.976248755,11.80867992,1.088135298,1.256936979
2.218305097,10.73544157,1.17096892,2.582973777
[k] is one of the parameters needed for runnning BNW, which set the number of high scoring networks to include in model averaging. The bigger k is, the longer the running time will be. Here, we set k = 20 as the default.
P0PCorNS will generate the JSD matrix file (inputPrefix_jsd_matrix.csv) and the list file of gene-JSD pairs (that is, inputPrefix_rearranged_jsd_matrix.csv), as well as the corresponding files for Gene_only mode(inputPrefix_jsd_matrix-GeneOnly.csv and inputPrefix_rearranged_jsd_matrix-GeneOnly.csv)
- demo_short_jsd_matrix.csv
rel,fmf,pp1,pna
0.05265369023718428,0.13325420573769514,0.052034339544552534,0.17669499131189442
0.1385596470513787,0.11302680891399997,0.09093054163610606,0.1947788583778019
0.0,0.14902906284970555,0.09922276826672846,0.13307726121938415
- demo_short_rearranged_jsd_matrix.csv
rel-1,0.05265369023718428
fmf-1,0.13325420573769514
pp1-1,0.052034339544552534
pna-1,0.17669499131189442
rel-2,0.1385596470513787
fmf-2,0.11302680891399997
pp1-2,0.09093054163610606
pna-2,0.1947788583778019
rel-3,0.0
fmf-3,0.14902906284970555
pp1-3,0.09922276826672846
pna-3,0.13307726121938415
- demo_short_jsd_matrix-GeneOnly.csv
rel,fmf,pp1,pna
0.17442418897700926,0.16267650468153744,0.10693559671088017,0.15127433096898724
0.12334827111012116,0.07317071878660704,0.052201967361136734,0.19480518170559802
5.705084205643186e-05,0.04079435020700513,0.14037564705260605,0.16052146322824934
- demo_short_rearranged_jsd_matrix-GeneOnly.csv
rel-1,0.17442418897700926
fmf-1,0.16267650468153744
pp1-1,0.10693559671088017
pna-1,0.15127433096898724
rel-2,0.12334827111012116
fmf-2,0.07317071878660704
pp1-2,0.052201967361136734
pna-2,0.19480518170559802
rel-3,5.705084205643186e-05
fmf-3,0.04079435020700513
pp1-3,0.14037564705260605
pna-3,0.16052146322824934