Table 2.
Whole-Genome Classification Confusion Matrix
| GS/Call | Class Prediction |
Average No. of IDC As a Result of Ties of Majority Votes | Total No. of Specimens in Every Class | CR | Sensitivity for Called Specimens | Specificity for Called Specimens | |||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| C1 | C2 | C3 | C4 | C5 | C6 | C7 | C8 | C9 | C10 | C11 | C12 | C13 | C14 | C15 | C16 | C17 | C18 | ||||||
| C1/GS | 9.3* | — | — | — | — | — | — | 0.7† | — | — | — | — | 0.3† | — | — | — | 0.3† | 1† | 1.3 | 13 | 0.897 | 0.800 | 0.999 |
| C2/GS | — | 70* | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 0 | 70 | 1.000 | 1.000 | 0.998 |
| C3/GS | — | — | 114.3* | — | — | — | 0.3† | 7.3† | — | — | — | — | — | — | — | — | — | — | 0 | 122 | 1.000 | 0.937 | 0.997 |
| C4/GS | — | — | — | 166.3* | — | — | — | 1.7† | — | — | — | — | 4† | — | — | — | 0.3† | 0.7† | 1 | 174 | 0.994 | 0.961 | 0.997 |
| C5/GS | — | — | — | — | 53.7* | — | — | 4.3† | — | — | — | — | — | — | — | — | — | — | 0 | 58 | 1.000 | 0.925 | 0.997 |
| C6/GS | — | 2† | — | — | — | 33* | — | 1† | — | — | — | — | — | — | — | — | — | — | 0 | 36 | 1.000 | 0.917 | 1.000 |
| C7/GS | — | — | — | — | — | — | 30.3* | 9.7† | — | — | — | — | — | — | — | — | — | — | 0 | 40 | 1.000 | 0.758 | 0.995 |
| C8/GS | 1† | 3† | 5.3† | 1† | 6† | — | 10.7† | 205.3* | — | — | — | — | 0.3† | 1† | 1† | — | 1.7† | — | 0.7 | 237 | 0.997 | 0.869 | 0.985 |
| C9/GS | — | — | — | — | — | — | — | — | 40* | — | — | — | — | — | — | — | — | — | 0 | 40 | 1.000 | 1.000 | 1.000 |
| C10/GS | — | — | — | — | — | — | — | — | — | 35* | — | — | 1† | — | — | — | 1† | — | 0 | 37 | 1.000 | 0.946 | 1.000 |
| C11/GS | — | — | — | — | — | — | — | — | — | — | 28* | — | — | — | — | — | — | — | 0 | 28 | 1.000 | 1.000 | 1.000 |
| C12/GS | — | — | — | 1† | — | — | — | — | — | — | — | 32* | 5† | — | — | — | — | — | 0 | 38 | 1.000 | 0.842 | 0.999 |
| C13/GS | 1† | — | — | 4.3† | — | — | — | 3† | — | 1† | — | 2† | 311.3* | 9.7† | 1.3† | 0.7† | 15.3† | — | 1.3 | 351 | 0.996 | 0.890 | 0.982 |
| C14/GS | — | — | — | — | — | — | — | — | — | — | — | — | 9.3† | 35.7* | — | — | 2.7† | — | 0.3 | 48 | 0.993 | 0.748 | 0.995 |
| C15/GS | — | — | — | — | — | — | — | — | — | — | — | — | 0.7† | — | 446* | — | 0.3† | — | 1 | 448 | 0.998 | 0.998 | 0.998 |
| C16/GS | — | — | 1† | — | — | — | — | — | — | — | — | — | — | — | — | 72* | 0.3† | 2.7† | 0 | 76 | 1.000 | 0.947 | 0.999 |
| C17/GS | — | — | — | — | — | — | — | — | — | — | — | — | 11† | — | — | — | 184.3* | 7.7† | 3 | 206 | 0.985 | 0.908 | 0.981 |
| C18/GS | — | — | — | — | — | — | — | 1† | — | — | — | — | — | — | 1† | 1† | 14† | 57* | 0 | 74 | 1.000 | 0.770 | 0.994 |
NOTE. Classification prediction results for 2,096 samples from stage I as analyzed by three 30-fold cross validations. Gold standard classes are given in rows C1/GS to C18/GS; the columns C1 to C18 list the average numbers of calls, rounded with up to 1 decimal place, for every class in three independent runs of cross-validations. This model uses trimmed mean of differences between perfect match and mismatch intensities with quantile normalization (DQN) signal intensities obtained from HG-U133 Plus 2.0 microarrays.
Abbreviations: IDC, indeterminable calls; CR, call rate [(No. of specimens–IDC)/No. of specimens]; GS, gold standard.
Values indicate correct prediction results.
Values represent misclassifications for each class.