Table 6.
Overview of the significant genes and proteins and the linear regression machine learning models. (A) List of the top six significant features from RNA sequencing and proteomics from TCGA Legacy datasets. (B) Model 1: An overview of the features used for the Mechanistic WNT pathway model. Models 2 and 3: An overview of the features used for the canonical WNT pathway, from CPTAC RNA and proteomics datasets, respectively. The number of features used and associated RMSE for each model is listed in the first column. “RMSE” = root mean squared error. Significant features are highlighted in blue.
| (A) Top Six Prognostic Genes and Proteins | Features |
|---|---|
|
mRNA Sequencing TCGA Legacy |
DVL3 |
| VANGL2 | |
| WNT6 | |
| TCF7L1 | |
| CER1 | |
| SOX17 | |
|
Proteomics TCGA Legacy |
PRKCA |
| ROCK2 | |
| CSNK2A1 | |
| LRP1 | |
| CSNK1A1 | |
| GPC4 | |
| (B) Linear Regression Models | Features |
|
Model 1: Mechanistic Model Model Input: CPTAC RNA RMSE: 1.0167 11 Features |
APC |
| AXIN1 | |
| CTNNB1 | |
| DKK2 | |
| DVL2 | |
| DVL3 | |
| GSK3B | |
| LRP6 | |
| SFRP1 | |
| SFRP2 | |
| WNT3 | |
|
Model 2: Canonical WNT model with top two prognostic genes and proteins Model Input: CPTAC RNA RMSE: 0.6416 9 Features |
DKK3 |
| FZD5 | |
| NKD1 | |
| NOTUM | |
| WNT11 | |
| DVL3 | |
| PRKCA | |
| ROCK2 | |
| VANGL2 | |
|
Model 3: Canonical WNT model with top two prognostic proteins Model Input: CPTAC proteomics RMSE: 1.0509 7 Features |
CTBP1 |
| CTBP2 | |
| GPC4 | |
| PLCB4 | |
| PRKCA | |
| RAC1 | |
| ROCK2 |