Python 3.6.3
numpy 1.14.2
sklearn 0.19.1
scipy 1.0.0
tensorflow 1.6.0
: https://www.dropbox.com/sh/tp70gitmmtaft0l/AABLSniRI9lCo1ZqtUGL4ZOqa?dl=0
:Comma-delimited file of gene expression data
Ex) ,patient1, patient2, patient3
gene1,-4.556,-1.784,2.295
gene2,-1.923,1.603,-2.696
:Comma-delimited file of copy number data
Ex) , patient1, patient2, patient3
gene1,-0.536,-0.464,8.025
gene2,7.022,-1.033,-0.636
:Comma-delimited file of DNA methylation data
Ex) , patient1, patient2, patient3
gene1,7.356,6.404,2.305
gene2,1.002,3.082,0.006
:Comma-delimited file of somatic mutation data
Ex) , patient 1, patient 2, patient 3
gene1,0,1,0
gene2,0,0,4
: Comma-delimited file of Patient's names and osevent
: Lable 0 = patient who has good prognosis
: Lable 1 = patient who has bad prognosis
Ex) patient1,0
patient2,1
patient3,0
: Comma-delimited file of functional interaction network
EX) GENE,GENE
gene1, gene2
gene1, gene3
gene4, gene5
python ProposedMethod.py [-t top_n_gene_in_ttest][-i n_experiment][-n n_gene][-d dampingfactor][-l limit_of_experiment] mRNA CNA METHYLATION SNP CLINICAL_FILE NETWORK
- top_n_gene_in_ttest : Parameter of step 1. Top N genes show statistical differences between good samples and poor samples in t-test. ( Default: 400 )
- n_experiment : Parameter of step 2 and step 3 (t in paper). To select a stable and robust feature for random initialization of weights, experiment t times repeatedly in GANs and PageRank step. ( Default : 5 )
- n_gene : Parameter of step 3 . The number of biomarkers selected for each experiment. ( Default: 250 )
- dampingfactor : Parameter of step 3. This is damping factor using in PageRank algorithm ( Default: 0.7 )
- limit_of_experiment : Parameter of step 2,3 (b in paper). when step 2 and step 3 are experimented t times repeatedly, the genes that appear b times in t experiment are selected as biomarkers. The b is the limit of experiment. ( Default: 5 )
python ProposedMethod.py BRCA_mRNA.txt BRCA_CNA.txt BRCA_methylation.txt BRCA_SNP.txt BRCA_Clinical.txt FIsnetwork.txt
1.preprocessing data...
loading data...
divide samples for 10fold validation
0fold ttest start
1fold ttest start
2fold ttest start
3fold ttest start
4fold ttest start
5fold ttest start
6fold ttest start
7fold ttest start
8fold ttest start
9fold ttest start
----------------------------------------------------------------------------------------------------
2. Step 1 : reconstructing FIs network
----------------------------------------------------------------------------------------------------
3. Step 2,3 : Learning the network and Feature selection using PageRank
start process process number : 0 fold number : 0
start process process number : 1 fold number : 0
start process process number : 2 fold number : 0
start process process number : 3 fold number : 0
start process process number : 4 fold number : 0
process 0 converge Epoch: 0002 n_iter : 0137 D_loss : -0.6936 G_loss : -0.6928
process 1 converge Epoch: 0002 n_iter : 0137 D_loss : -0.6937 G_loss : -0.6926
process 3 converge Epoch: 0002 n_iter : 0137 D_loss : -0.6934 G_loss : -0.693
process 2 converge Epoch: 0002 n_iter : 0137 D_loss : -0.6935 G_loss : -0.6928
process 4 converge Epoch: 0002 n_iter : 0137 D_loss : -0.6937 G_loss : -0.6928
start process process number : 0 fold number : 1
start process process number : 1 fold number : 1
start process process number : 2 fold number : 1
start process process number : 3 fold number : 1
start process process number : 4 fold number : 1
process 4 converge Epoch: 0002 n_iter : 0138 D_loss : -0.6934 G_loss : -0.6933
process 3 converge Epoch: 0002 n_iter : 0138 D_loss : -0.6933 G_loss : -0.6931
process 2 converge Epoch: 0002 n_iter : 0138 D_loss : -0.6932 G_loss : -0.6934
process 0 converge Epoch: 0002 n_iter : 0138 D_loss : -0.6935 G_loss : -0.693
process 1 converge Epoch: 0002 n_iter : 0138 D_loss : -0.6934 G_loss : -0.6931
start process process number : 0 fold number : 2
start process process number : 1 fold number : 2
start process process number : 2 fold number : 2
start process process number : 3 fold number : 2
start process process number : 4 fold number : 2
process 1 converge Epoch: 0002 n_iter : 0138 D_loss : -0.6932 G_loss : -0.6932
process 2 converge Epoch: 0002 n_iter : 0138 D_loss : -0.6934 G_loss : -0.693
process 4 converge Epoch: 0002 n_iter : 0138 D_loss : -0.6935 G_loss : -0.693
process 0 converge Epoch: 0002 n_iter : 0138 D_loss : -0.693 G_loss : -0.6934
process 3 converge Epoch: 0002 n_iter : 0138 D_loss : -0.6934 G_loss : -0.693
start process process number : 0 fold number : 3
start process process number : 1 fold number : 3
start process process number : 2 fold number : 3
start process process number : 3 fold number : 3
start process process number : 4 fold number : 3
process 0 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6931 G_loss : -0.6932
process 1 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6926 G_loss : -0.6937
process 2 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6926 G_loss : -0.6937
process 3 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6927 G_loss : -0.6936
process 4 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6931 G_loss : -0.6932
start process process number : 0 fold number : 4
start process process number : 1 fold number : 4
start process process number : 2 fold number : 4
start process process number : 3 fold number : 4
start process process number : 4 fold number : 4
process 3 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6922 G_loss : -0.6941
process 4 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6928 G_loss : -0.6936
process 2 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6927 G_loss : -0.6936
process 0 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6921 G_loss : -0.6942
process 1 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6925 G_loss : -0.6939
start process process number : 0 fold number : 5
start process process number : 1 fold number : 5
start process process number : 2 fold number : 5
start process process number : 3 fold number : 5
start process process number : 4 fold number : 5
process 1 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6937 G_loss : -0.6927
process 2 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6935 G_loss : -0.6928
process 0 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6931 G_loss : -0.6932
process 4 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6932 G_loss : -0.6931
process 3 converge Epoch: 0002 n_iter : 0139 D_loss : -0.693 G_loss : -0.6934
start process process number : 0 fold number : 6
start process process number : 1 fold number : 6
start process process number : 2 fold number : 6
start process process number : 3 fold number : 6
start process process number : 4 fold number : 6
process 2 converge Epoch: 0002 n_iter : 0139 D_loss : -0.692 G_loss : -0.6943
process 4 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6927 G_loss : -0.6936
process 1 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6923 G_loss : -0.694
process 0 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6924 G_loss : -0.6939
process 3 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6923 G_loss : -0.694
start process process number : 0 fold number : 7
start process process number : 1 fold number : 7
start process process number : 2 fold number : 7
start process process number : 3 fold number : 7
start process process number : 4 fold number : 7
process 0 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6923 G_loss : -0.6941
process 3 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6922 G_loss : -0.6942
process 4 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6929 G_loss : -0.6935
process 1 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6927 G_loss : -0.6937
process 2 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6924 G_loss : -0.6939
start process process number : 0 fold number : 8
start process process number : 1 fold number : 8
start process process number : 2 fold number : 8
start process process number : 3 fold number : 8
start process process number : 4 fold number : 8
process 3 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6923 G_loss : -0.694
process 2 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6932 G_loss : -0.6931
process 4 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6923 G_loss : -0.694
process 0 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6923 G_loss : -0.694
process 1 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6928 G_loss : -0.6936
start process process number : 0 fold number : 9
start process process number : 1 fold number : 9
start process process number : 2 fold number : 9
start process process number : 3 fold number : 9
start process process number : 4 fold number : 9
process 4 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6934 G_loss : -0.6929
process 0 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6935 G_loss : -0.6928
process 1 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6936 G_loss : -0.6928
process 3 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6931 G_loss : -0.6932
process 2 converge Epoch: 0002 n_iter : 0139 D_loss : -0.6935 G_loss : -0.6928
----------------------------------------------------------------------------------------------------
4. Step4 : Prognosis Prediction
layer= [5] alpha= 100 10fold AUC= 0.7397354497354498
layer= [5] alpha= 150 10fold AUC= 0.7485714285714286
layer= [5] alpha= 200 10fold AUC= 0.737989417989418
layer= [10] alpha= 100 10fold AUC= 0.7373544973544973
layer= [10] alpha= 150 10fold AUC= 0.752962962962963
layer= [10] alpha= 200 10fold AUC= 0.7476719576719577
- File name: ProposedMethod_biomarker_per_fold.txt
- Selected biomarkers of each fold.