Figure 2

Probabilistic coevolutionary biclustering algorithm. Pop(G) is a population for gene set and Pop(C) is that for condition set. Individuals, x _i and y _j are evaluated and the bests are selected. The probability vectors of two populations, P _G and P _C are updated and new populations are generated by sampling and mutation in each iteration. Each parameter indicates: δ (cutoff of residue score); μ and ν (initial size of gene and condition population); w _b and w _v (parameters controlling the variance and volume); w _g and w _c (parameters keeping a balance between the genes and condition); α and β (parameters controlling update of probability); S _g (best individuals in genes); S _c (best individuals in conditions, respectively).