Table 4.
Prediction on archived GOA files using BMA (see Eq. (6)) by combining the GO annotations in CC, MF and BP together and then evaluating in each sub-ontology
| MicroAvgF1 | MacroAvgF1 | 1-RankLoss | Fmax | RAccuracy | ||
|---|---|---|---|---|---|---|
| CC | H →H | 0.8700 | 0.4416 | 0.9682 | 0.8619 | 0.2057 |
| M →H | 0.8550 | 0.4407 | 0.9310 | 0.8551 | 0.1963 | |
| M+H →H | 0.8765 | 0.4451 | 0.9791 | 0.9006 | 0.2457 | |
| D →H | 0.8543 | 0.4372 | 0.9387 | 0.8610 | 0.1626 | |
| D+H →H | 0.8666 | 0.4412 | 0.9652 | 0.8773 | 0.1852 | |
| A →H | 0.8424 | 0.4388 | 0.8862 | 0.8595 | 0.1428 | |
| A+H →H | 0.8673 | 0.4358 | 0.9518 | 0.8761 | 0.1895 | |
| M →M | 0.8193 | 0.4430 | 0.9487 | 0.8481 | 0.1556 | |
| H →M | 0.8155 | 0.4416 | 0.9475 | 0.8514 | 0.1582 | |
| H+M →M | 0.8256 | 0.4507 | 0.9692 | 0.8795 | 0.1853 | |
| D →M | 0.8085 | 0.4433 | 0.9289 | 0.8490 | 0.1446 | |
| D+M →M | 0.8170 | 0.4474 | 0.9460 | 0.8560 | 0.1452 | |
| A →M | 0.7963 | 0.4258 | 0.9121 | 0.8160 | 0.1157 | |
| A+M →M | 0.8162 | 0.4377 | 0.9241 | 0.8385 | 0.1410 | |
| MF | H →H | 0.8539 | 0.4287 | 0.9569 | 0.8394 | 0.1983 |
| M →H | 0.8514 | 0.4282 | 0.9468 | 0.8352 | 0.1721 | |
| M+H →H | 0.8606 | 0.4312 | 0.9721 | 0.8785 | 0.2349 | |
| D →H | 0.8513 | 0.4232 | 0.9507 | 0.8290 | 0.1358 | |
| D+H →H | 0.8532 | 0.4294 | 0.9540 | 0.8451 | 0.1945 | |
| A →H | 0.8435 | 0.4217 | 0.9060 | 0.8049 | 0.0921 | |
| A+H →H | 0.8453 | 0.4239 | 0.9394 | 0.8187 | 0.1508 | |
| M →M | 0.7980 | 0.4015 | 0.9426 | 0.8066 | 0.1528 | |
| H →M | 0.7963 | 0.3927 | 0.9246 | 0.8001 | 0.1501 | |
| H+M →M | 0.8078 | 0.4090 | 0.9672 | 0.8629 | 0.1937 | |
| D →M | 0.7596 | 0.3936 | 0.9096 | 0.7748 | 0.1108 | |
| D+M →M | 0.7989 | 0.4053 | 0.9427 | 0.8216 | 0.1563 | |
| A →M | 0.7452 | 0.3883 | 0.8856 | 0.7528 | 0.0815 | |
| A+M →M | 0.7949 | 0.3984 | 0.9274 | 0.7829 | 0.1395 | |
| BP | H →H | 0.8376 | 0.7977 | 0.9522 | 0.8023 | 0.2058 |
| M →H | 0.8320 | 0.7861 | 0.9267 | 0.8134 | 0.1791 | |
| M+H →H | 0.8450 | 0.8055 | 0.9694 | 0.8390 | 0.2421 | |
| D →H | 0.8374 | 0.7917 | 0.9513 | 0.8041 | 0.1948 | |
| D+H →H | 0.8370 | 0.7978 | 0.9502 | 0.8098 | 0.2032 | |
| A →H | 0.8248 | 0.7840 | 0.8998 | 0.8119 | 0.1433 | |
| A+H →H | 0.8322 | 0.7830 | 0.9328 | 0.7941 | 0.1796 | |
| M →M | 0.7814 | 0.6968 | 0.9372 | 0.7916 | 0.1864 | |
| H →M | 0.7892 | 0.6884 | 0.9384 | 0.7901 | 0.1897 | |
| H+M →M | 0.7949 | 0.7132 | 0.9611 | 0.8276 | 0.2363 | |
| D →M | 0.7829 | 0.6999 | 0.9364 | 0.7853 | 0.1818 | |
| D+M →M | 0.7820 | 0.7033 | 0.9365 | 0.7910 | 0.1883 | |
| A →M | 0.7694 | 0.6897 | 0.9023 | 0.7769 | 0.1417 | |
| A+M →M | 0.7779 | 0.6874 | 0.9199 | 0.7822 | 0.1732 |
H →H directly uses GO annotations of Human to predict annotations of Human genes. M →H only employs annotations of genes from Mouse to predict annotations of Human genes. M+H →H uses GO annotations of genes from Mouse and Human to predict annotations of Human genes. D+H →H uses annotations of genes from Danio rerio and Human to predict annotations of Human genes. A+H →H uses annotations of genes from Arabidopsis thaliana and Human to predict annotations of Human genes. M →M, H+M →M, D+M →M and A+M →M follow the similar protocol, but predict annotations of Mouse genes. The data in boldface is the statistically significant best among these comparing methods for a particular target species, and the significance is checked by paired t-test at 95% confidence intervals