Table 1.
LLE-DML |
PCA-DML |
LLE-ED |
PCA-ED |
Euclidean distance |
Drug–disease score Sirota et al. (2011) |
Anti-correlation |
|||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Dataset | NDCG | P-value | NDCG | P-value | NDCG | P-value | NDCG | P-value | NDCG | P-value | NDCG | P-value | NDCG | P-value | |
Breast cancer | GSE26910 | 0.3348 | 0.0140 | 0.1586 | 0.2398 | 0.1317 | 0.5904 | 0.1486 | 0.3207 | 0.1820 | 0.1299 | 0.2533 | 0.0429 | 0.1827 | 0.1239 |
GSE1299 | 0.3355 | 0.0140 | 0.2695 | 0.0329 | 0.1421 | 0.3956 | 0.1741 | 0.1648 | 0.2095 | 0.0779 | 0.1286 | 0.6783 | 0.2096 | 0.0779 | |
TCGA-BRCA | 0.2443 | 0.0480 | 0.1465 | 0.3387 | 0.1482 | 0.3237 | 0.1585 | 0.2398 | 0.1538 | 0.2747 | 0.1523 | 0.2847 | 0.1517 | 0.2937 | |
Idiopathic pulmonary fibrosis | LGRC | 0.2445 | 0.0460 | 0.1609 | 0.3037 | 0.1404 | 0.5894 | 0.1611 | 0.2987 | 0.1652 | 0.2507 | 0.1437 | 0.5325 | 0.1562 | 0.3497 |
GSE33577 | 0.3639 | 0.0180 | 0.1730 | 0.2088 | 0.1614 | 0.2987 | 0.1609 | 0.3037 | 0.1588 | 0.3247 | 0.1566 | 0.3447 | 0.1560 | 0.3526 | |
GSE55384 | 0.2468 | 0.0410 | 0.1871 | 0.1528 | 0.1493 | 0.4396 | 0.1671 | 0.2378 | 0.1695 | 0.2238 | 0.1408 | 0.5844 | 0.1654 | 0.2507 | |
GSE24206 | 0.2579 | 0.0350 | 0.1339 | 0.7572 | 0.1654 | 0.2488 | 0.1934 | 0.1339 | 0.2095 | 0.0899 | 0.2511 | 0.0370 | 0.1719 | 0.2138 | |
Rheumatoid arthritis | GSE1919 | 0.1955 | 0.1069 | 0.1324 | 0.5554 | 0.1819 | 0.1339 | 0.1337 | 0.5285 | 0.1342 | 0.5155 | 0.1452 | 0.3516 | 0.1377 | 0.4625 |
GSE15573 | 0.1862 | 0.1209 | 0.1454 | 0.3497 | 0.1474 | 0.3287 | 0.1427 | 0.3846 | 0.1704 | 0.1738 | 0.1358 | 0.4835 | 0.1655 | 0.1928 |
Note: The NDCG value is determined to be significant based on the bootstrap procedure. Rows with bold text show the highest and the most significant (P-value <0.05) NDCG values. As shown, the LLE-DML method performs better in terms of placing the FDA-approved drugs for their approved indications at the top of their ranked list.