Table 1.
ARIs of four feature selection methods combined with four clustering methods across 12 datasets
| Feature selection + clustering method |
Datasets* (Details presented in Additional file1– Public datasets) |
|||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Pr | C | Mn | R1 | R2 | NPh | Lm | R3 | B | T | Br | L | |
| COPA+CH |
0.12 |
0.04 |
0.69 |
0.20 |
0.64 |
0.15 |
0.23 |
0.38 |
0.06 |
0.05 |
0.16 |
0.45 |
| COPA+KM |
0.30 |
0.16 |
0.53 |
0.62 |
0.33 |
0.31 |
0.25 |
0.54 |
0.09 |
0.23 |
0.12 |
0.41 |
| COPA+PAM |
0.13 |
0.18 |
0.60 |
0.90 |
0.81 |
0.36 |
0.26 |
0.57 |
−0.02 |
0.31 |
0.12 |
0.35 |
| COPA+SIL |
0.04 |
0.08 |
0.69 |
0.20 |
0.30 |
0.15 |
0.33 |
0.43 |
0.06 |
0.05 |
0.07 |
0.55 |
| DE+CH |
0.17 |
0.15 |
0.21 |
0.24 |
0.53 |
0.36 |
0.38 |
0.28 |
0.21 |
0.52 |
0.12 |
0.44 |
| DE+KM |
0.29 |
0.15 |
0.51 |
0.75 |
0.59 |
0.34 |
0.38 |
0.65 |
0.27 |
0.63 |
0.12 |
0.54 |
| DE+PAM |
0.35 |
0.13 |
0.24 |
0.79 |
0.76 |
0.34 |
0.26 |
0.56 |
0.11 |
0.46 |
0.16 |
0.43 |
| DE+SIL |
0.17 |
0.15 |
0.33 |
0.24 |
0.53 |
0.14 |
0.38 |
0.28 |
0.21 |
0.52 |
0.12 |
0.44 |
| mCOPA+CH |
0.29 |
0.15 |
0.60 |
0.55 |
0.40 |
0.35 |
0.38 |
0.39 |
0.06 |
0.52 |
0.11 |
0.30 |
| mCOPA+KM |
0.46 |
0.01 |
0.79 |
0.68 |
0.48 |
0.36 |
0.45 |
0.85 |
0.00 |
0.63 |
0.08 |
0.62 |
| mCOPA+PAM |
0.47 |
0.10 |
0.55 |
0.82 |
0.54 |
0.45 |
0.44 |
0.49 |
0.03 |
0.50 |
0.20 |
0.44 |
| mCOPA+SIL |
0.29 |
0.14 |
0.60 |
0.61 |
0.40 |
0.35 |
0.38 |
0.39 |
0.06 |
0.52 |
0.11 |
0.30 |
| VAR+CH |
0.14 |
0.08 |
0.69 |
0.32 |
0.34 |
0.28 |
0.38 |
0.57 |
0.02 |
0.26 |
0.13 |
0.45 |
| VAR+KM |
0.16 |
0.16 |
0.47 |
0.81 |
0.47 |
0.23 |
0.34 |
0.64 |
0.09 |
0.15 |
0.17 |
0.41 |
| VAR+PAM |
0.17 |
0.16 |
0.88 |
0.81 |
0.43 |
0.02 |
0.26 |
0.61 |
0.10 |
0.06 |
0.15 |
0.33 |
| VAR+SIL | 0.14 | 0.08 | 0.69 | 0.89 | 0.34 | 0.28 | 0.38 | 0.57 | 0.02 | 0.12 | 0.13 | 0.59 |
Note: the ARI scores in italicized bold indicate the best performing method for each dataset. Datasets in bold indicate those in which mCOPA provided the most informative feature selection for the clustering of clinical subtypes.
*Datasets: Pr (Prostate: GSE6099); C (Cervical: GSE7410); Mn (Melanoma: GSE7553); R1 (Renal: GSE11024); R2 (Renal: GSE11151); NPh (Nasopharangeal: GSE12452); Lm (Lymphoma: GSE12453); R3 (Renal: GSE15641); B (Brain: GSE15824); T (Thyroid: GSE29265); Br (Breast: GSE29431); L (Lung: GSE32036).