Table 1 Protein complex detection algorithm

Input : an unweighted network, a weighted network built via GO annotation and a training set

Complex detection process:

Step 1: construct the feature vector space for the complexes in the training set from the unweighted and weighted PIN networks and train the Regression model

Step 2: find maximal cliques in the PIN by the Cliques algorithm

    -rank the clique set C={C₁, C₂, ..., C_n} in descending order of the scores given by the Regression model

    -for each clique C_i, check all the cliques (denoted as C_j) with lower scores, if C_i∩C_j > threshold, then remove C_j.

    -output: the updated clique set

Step 3: grow the cliques

    -for each clique C_i, the set of its neighbors is denoted as N(C_i), do update operation as follows:

        -check all the nodes in N(C_i)

        -select v_i∈N(C_i), which makes v_i∪C_i achieve higher score given by the Regression model

        -update C_i= v_i∪C_i, N(C_i) = N(C_i) - v_i

            -repeat the update operation until there is no node v_j in N(C_i) that leads to score(v_j∪C_i) > score(C_i)

        -output: the candidate complex set C = {C₁, C₂, ..., C_n}

Step 4: filter the candidate complexes

    -rank the candidate complexes in descending order of the score given by the Regression model

    -for each candidate complex C_i, check all the candidates C_j with lower scores

        -if overlap (C_i, C_j) > merg_thred

            if score(C_i∪C_j) > score(C_i) do merge operation: update C_i = C_i∪C_j

            else do remove operation: remove C_j from the candidate set

output: the predicted complex set