Table 2.
Bias due to the order of gene expression values.
| Class/Range | C1 | C2 | C3 | Class/Range | C1 | C2 | C3 |
|---|---|---|---|---|---|---|---|
| R1 | 3 | 0 | 0 | R1 | 3 | 0 | 0 |
| R2 | 0 | 88 | 0 | R2 | 4 | 0 | 0 |
| R3 | 4 | 0 | 0 | R3 | 0 | 88 | 0 |
| R4 | 0 | 0 | 5 | R4 | 0 | 0 | 5 |
In this example, we consider the goodness ranking of two genes in a dataset with 100 patients and 3 classes of cancer sub-types. The expression values of the genes were discretized into 4 ranges R1, R2, R3 and R4. There are 4 patients in class C1 with expression values of gene #1 in the range R3, 88 patients in class C2 with expression values of gene #1 in the range R2, and so on.