Skip to main content

Table 4 Classifier performance in predicting GO terms using individual sources of data and some of their combinations using only data from mouse.

From: Combining heterogeneous data sources for accurate functional annotation of proteins

Source   AUC    P20R  
  MF BP CC MF BP CC
BLAST 0.77 0.61 0.69 0.40 0.13 0.25
Sequence 0.83 0.65 0.76 0.41 0.14 0.26
PPI 0.78 0.80 0.81 0.33 0.25 0.43
Protein-GO term co-mention 0.78 0.75 0.79 0.24 0.17 0.33
Expression 0.58 0.64 0.62 0.04 0.06 0.10
PPI + co-mention 0.85 0.82 0.85 0.43 0.29 0.45
PPI + co-mention + expression 0.86 0.83 0.86 0.42 0.29 0.46
  1. BLAST refers to a classifier trained on BLAST scores only; the Sequence entry uses all the sequence-based features. In addition to classifiers trained on PPI, co-mention and expression individually, we also provide results using PPI and co-mention and the combination of all three.