Algorithm 1.

The updated process of CMFMTL.

Input: known drug-disease therapeutic association matrix,          Ap ∈ {0, 1}m×n;          known drug-disease non-therapeutic association matrix,          An ∈ {0, 1}m×n;          drug similarity matrix, wr ∈ ℝm×m;          disease similarity matrix, wd ∈ ℝn×n;          dimensionality of the embedded space, k < min(m, n);          regularization parameters, α > 0, β > 0 and λ> 0 Output: the prediction matrices Ap*, An* Initialize V ∈ ℝn×k and U ∈ ℝm×k in the interval [0, 1] randomly; Y = 0 and Z = 0;          ρ1 = ρ2 = 1 Repeat          Update Rp and Rn using                                       Rp = CG(Rp), Rn = CG(Rn)          Update J, W, U and V via the equation (11)          Update Y, Z, ρ1 and ρ2 via the equation (13) End until convergence Output Ap*, An* using                                       Ap* = URpVT,An* = URnVT