Skip to main content

Table 2 Summary of network-based features

From: Protein-protein interaction prediction using a hybrid feature representation and a stacked generalization scheme

Features

Definitiona

Common neighbors

|N(Pi) ∩ N(Pj)|

Jaccard index

\( \frac{\left|N\left({P}_i\right)\cap N\left({P}_j\right)\right|}{\left|N\left({P}_i\right)\cup N\left({P}_j\right)\right|} \)

Adamic–Adar index

\( \sum \limits_{P_k\in N\left({P}_i\right)\cap N\left({P}_j\right)}\frac{1}{\log \left|N\left({P}_k\right)\right|} \)

Preferential attachment score

|N(Pi)| × |N(Pj)|

Otsuka–Ochiai coefficient

\( \frac{\left|N\left({P}_i\right)\cap N\left({P}_j\right)\right|}{\sqrt{\left|N\left({P}_i\right)\times N\left({P}_j\right)\right|}} \)

  1. aN(Pi) denotes the set of Pi’s neighbors