Skip to main content

Table 7 Analysis of prediction results by the gene ontology slims

From: Critical assessment of sequence-based protein-protein interaction prediction methods that do not require homologous protein sequences

Results on the yeast data sorted according to AUC

 

GO term

# of cases

Best method

AUC

GO term explanation

1

0005198

39513

C1

0.90

Structural molecular activity

2

0007124

9192

C

0.89

Pseudohyphal growth

3

0006997

10093

C

0.89

Nucleus organization

4

0007047

18668

C

0.89

Cell wall organization

5

0005215

44019

C

0.89

Transporter activity

Results on the yeast data sorted according to P20R

 

GO term

# of cases

Best method

P20R

GO term explanation

1

0005618

8689

M2

1.00

Cell wall

2

0006997

10093

C

0.97

Nucleus organization

3

0042254

44304

C

0.95

Ribosome biogenesis

4

0005198

39513

C

0.92

Structural molecule activity

5

0008289

10690

M2

0.92

Lipid binding

Results on the human data sorted according to AUC

 

GO term

# of cases

Best method

AUC

GO term explanation

1

0008907

245

C

1.00

Integrase activity

2

0004871

71939

C

0.92

Signal transducer activity

3

0051704

88280

C

0.92

Multi-organism process

4

0008219

98990

C

0.92

Cell death

5

0016740

244001

C

0.91

Transferase activity

Results on the human data sorted according to P20R

 

GO term

# of cases

Best method

P20R

GO term explanation

1

0009405

1017

M2

1.00

Pathogenesis

2

0008907

245

M2

1.00

Integrase activity

3

0004871

71939

C

0.91

Signal transducer activity

4

0004872

208752

C

0.88

Receptor activity

5

0016301

110554

C

0.88

Kinase activity

Results on the combined data sorted according to AUC

 

GO term

# of cases

Best method

AUC

GO term explanation

1

0008907

245

C

0.99

Integrase activity

2

0004871

77553

C

0.92

Signal transducer activity

3

0015267

7183

C

0.91

Channel activity

4

0004872

208752

C

0.91

Receptor activity

5

0051704

88280

C

0.91

Multi-organism process

Results on the combined data sorted according to P20R

 

GO term

# of cases

Best method

P20R

GO term explanation

1

0005618

8689

M2

1.00

Cell wall

2

0009405

1017

M2

1.00

Pathogenesis

3

0008907

245

M2

1.00

Integrase activity

4

0006997

10093

M2

0.97

Nucleus organization

5

0008289

10690

M2

0.92

Lipid binding

  1. 1C: the consensus method that integrates the four methods M1 through M4.
  2. For each combination of a data set (the yeast, the human or the combined data) and an evaluation scheme (AUC or P20R), five GO terms are listed for which best performance was achieved. For each GO term, the number of protein-protein pairs in the data set is shown in the third column for which either protein in the pair is annotated with that GO term. Also shown are the best-performing method (column 4) and its performance (column 5).
Back to article page

BMC Bioinformatics

ISSN: 1471-2105

Contact us