Skip to main content
Fig. 2 | BMC Bioinformatics

Fig. 2

From: Assigning protein function from domain-function associations using DomFun

Fig. 2

DomFun evaluation in terms of maximum F-measure (\(F_{max}\)) calculation when predicting for Gene Ontology (GO) molecular function (GOMF) (a), biological process (GOBP) (b) and cellular component (c) (GOCC) terms using the CAFA 3 prediction benchmark. The associations between GO terms and protein domains, classified using FunFams (FF) and superfamilies (SF) separately, were calculated using four different association indices: Jaccard (Jac), Simpson (Sim), Pearson correlation coefficient (PCC) and hypergeometric index (HyI), and combined using either Fisher’s method (Fis) or Stouffer’s method (Sto). Results for the baseline methods BLAST and Naïve are also included for comparison. Coverage (C) values for each method are included within the bars. The CAFA 3 evaluation procedure was set to partial mode and limited knowledge

Back to article page