Skip to main content

Table 3 Percentages of the top 500 predicted annotations found confirmed for each method (LSI, SIM, pLSA), weighting schema (none, NTN, NTM, ATN), and dataset (Bos taurus, Danio rerio, Drosophila melanogaster).

From: Computational algorithms to predict Gene Ontology annotations

 

Bos Taurus

Danio rerio

Drosophila m.

Method

cmp

Uany

Ucur

cmp

Uany

Ucur

cmp

Uany

Ucur

LSI-none

26.4

26

3.6

13

21.2

11.6

31.2

30

6.8

LSI-NTN

22.2

24.8

7.2

9.4

27.4

16.8

6.6

13.8

7.2

LSI-NTM

14.8

19.2

6.4

6.6

17.6

11.6

9.8

26

17

LSI-ATN

21.6

28.2

9.4

6.2

27.6

22.6

23.6

24

5.8

SIM-none

19.2

19

4.4

12.4

21.8

10.8

32.6

35

10.6

SIM-NTN

17.4

20.6

7

11.4

28.8

17.8

7.4

24.6

17

SIM-NTM

22

24

6.2

7.4

30.2

21.4

16

45.6

35.2

SIM-ATN

24

32.2

10.4

6

31.6

26.8

22.6

23.8

6

pLSA-none

27.4

20.6

5.2

14

24.4

13

4.8

6.6

4.2

pLSA-NTN

19.6

21.2

6.8

11.8

21.6

13.2

8

11.6

5.6

pLSA-NTM

14.6

20.2

5

15.8

23.8

11.6

5.4

7

3.2

pLSA-ATN

15.6

16.2

6.4

4.8

9.6

5

3.8

6.4

4

  1. We report: the portion of predictions that were confirmed in the outdated dataset with only computational evidence, therefore not included in the input corpus (cmp); the portion of predictions that were confirmed in the updated dataset with any evidence (Uany) and the portion of confirmed predictions in the updated dataset with not-computational evidence (Ucur). The best values obtained for each of these three cases and dataset across the considered methods are highlighted in bold.