TABLE 3.
Distribution of class predictions for test set samples
| Actual identity of unknown sample | Predicted class*
|
|||||
|---|---|---|---|---|---|---|
| LADCA | LSCC | SCLC | PC | NL | No call | |
| LADCA | 12 | 1 | ||||
| LSCC | 1 | 12 | ||||
| PR | 24 | |||||
| CO | 19 | 1 | ||||
| BR | 20 | 1 | 1 | 3 | ||
| GA | 9 | 1 | 1 | 1 | ||
| PA | 4 | 1 | 1 | |||
The value in each box is the number of primary tumors from the test set (n = 113) predicted with a given identity from the expression ratio-based model developed in the training set of samples. LADCA, Lung adenocarcinoma; LSCC, lung SCC; PC, pulmonary carcinoid; NL, normal lung; PR, prostate adenocarcinomas; CO, colon adenocarcinomas; BR, breast adenocarcinomas; GA, gastroesophageal adenocarcinomas; PA, pancreatic adenocarcinomas.