Table 6 Incorporating covariates into the model. Models treating the fitting of counts for tag GCGAAACCCT from Table 2, with the cell lines hypothetically allocated to normal tissue B (libraries 5 and 6) and cancer tissue B (libraries 7 and 8). This division is made to illustrate how the effects of two differences, normal vs cancer and tissue A vs tissue B (β1 and β2 respectively) can be partitioned according to their importance. In Model 2, we have further introduced a continuous covariate (β3) corresponding to the levels of a biomarker to show how that can be figured in as well.
From: Overdispersed logistic regression for SAGE: Modelling multiple groups and covariates
Model 1: | Hypothetical Cov. |
| df = 5 | |
|---|---|---|---|---|
Coefficients | Estimate | (s.e) | t-value | p-value |
β 0 | -4.928 | 0.291 | -16.921 | 1.318e - 05 |
β 1 | -1.293 | 0.593 | -2.181 | 0.0810 |
β 2 | -1.956 | 0.738 | -2.650 | 0.0454 |
Model 2: | Hypothetical Biom. |
| df = 4 | |
Coefficients | Estimate | (s.e) | t-value | p-value |
β 0 | -4.167 | 0.608 | -6.851 | 1.012e-03 |
β 1 | -1.423 | 0.611 | -2.328 | 0.0674 |
β 2 | -2.031 | 0.752 | -2.700 | 0.0428 |
β 3 | -1.365 | 1.028 | -1.328 | 0.2417 |
= 1.224e - 03