Supporting text, tables and figures.

1. Supplementary Figs: Fig S1: Sizes of the subnetworks identified by PinnacleZ in the experiment using the network PPI_1 and the simulated data with normal distribution. Fig S2: Sizes of the subnetworks identified by PinnacleZ in the experiment using the network PPI_2 and the sampled data from RNA-Seq TCGA breast cancer dataset. Fig S3: Sizes of the subnetworks identified by COSINE in the experiment using the network PPI_1 and the simulated data with normal distribution. Fig S4: Sizes of the subnetworks identified by COSINE in the experiment using the network PPI_2 and the sampled data from RNA-Seq TCGA breast cancer dataset. Fig S5: Sizes of the subnetworks identified by jActiveModules in the experiment using the network PPI_1 and the simulated data with normal distribution. Fig S6: Sizes of the subnetworks identified by jActiveModules in the experiment using the network PPI_2 and the sampled data from RNA-Seq TCGA breast cancer dataset. Fig S7: Sizes of the subnetworks identified by all the methods in the experiment using the network PPI_1 and the simulated data with normal distribution. Fig S8: Sizes of the subnetworks identified by all the methods in the experiment using the network PPI_2 and the sampled data from RNA-Seq TCGA breast cancer dataset. Fig S9: Yao’s dataset, biopsies: Active modules 1–18. Fig S10: Yao’s dataset, myoblasts: Active modules 1–10. Fig S11: Yao’s dataset, myotubes: Active modules 1–23. Fig S12: Banerji’s 2017 dataset: Active modules 1–23. Fig S13: Banerji’s 2019 dataset: Active modules 1–17. 2. Supplementary Tables. Table S1: Samples from Yao’s datasets. Downloaded from https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE56787. Table S2: Samples from Banerji’s 2017 dataset. Downloaded from https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE102812. Table S3: Samples from Banerji’s 2019 dataset. Downloaded from https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE123468. 3. Non-dominated Sorting Genetic Algorithm II (NSGA-II). Algorithm S1: Fast non-dominated sorting. Fig S14: Crowding distance concept. Algorithm S2: Crowding distance assignment. Algorithm S3: Elitist selection of a new population. 4. MOGAMUN Genetic Algorithm parameter tuning. Fig S15: Convergence plots of the average nodes score (A) and density (B). At each generation, the best values for the average nodes score and density of the 30 runs are averaged and plotted. Fig S16: Average nodes score (A), density (B), and overlapping nodes (C) of the active modules obtained in the accumulated Pareto fronts of 30 runs of MOGAMUN with different combinations of parameters. 5. Application to Facio-Scapulo-Humeral muscular Dystrophy type 1 (FSHD1). Fig S17: Sizes of the subnetworks identified by the five approaches on Yao’s dataset, biopsies. Fig S18: Size, average nodes score and density of the subnetworks obtained by the different methods using Yao’s dataset, biopsies. The sizes of the subnetworks are represented on a log scale. The density is computed either on the aggregated network, corresponding to the union of the three biological networks used in this study, or using the multiplex-normalized density, proposed in the main manuscript. Fig S19: Size, average nodes score and density of the subnetworks obtained by the different methods using Yao’s dataset, biopsies, selecting only the subnetworks containing at least 15 nodes. The sizes of the subnetworks are represented on a log scale. The density is computed either on the aggregated network, corresponding to the union of the three biological networks used in this study, or using the multiplex-normalized density, proposed in the main manuscript. Fig S20: Sizes of the subnetworks identified by the five approaches on Yao’s dataset, myotubes. Fig S21: Size, average nodes score and density of the subnetworks obtained by the different methods using Yao’s dataset, myotubes. The sizes of the subnetworks are represented on a log scale. The density is computed either on the aggregated network, corresponding to the union of the three biological networks used in this study, or using the multiplex-normalized density, proposed in the main manuscript. Fig S22: Size, average nodes score and density of the modules obtained by the different methods using Yao’s dataset, myotubes, selecting only the subnetworks containing at least 15 nodes. The sizes of the subnetworks are represented on a log scale. The density is computed either on the aggregated network, corresponding to the union of the three biological networks used in this study, or using the multiplex-normalized density, proposed in the main manuscript. Fig S23: Sizes of the subnetworks identified by the five approaches on Yao’s dataset, myoblasts. Fig S24: Size, average nodes score and density of the subnetworks obtained by the different methods, using Yao’s dataset, myoblasts. The sizes of the subnetworks are represented on a log scale. The density is computed either on the aggregated network, corresponding to the union of the three biological networks used in this study, or using the multiplex-normalized density, proposed in the main manuscript. Fig S25: Size, average nodes score and density of the subnetworks obtained by the different methods, using Yao’s dataset, myoblasts, selecting only the modules containing at least 15 nodes. The sizes of the subnetworks are represented on a log scale. The density is computed either on the aggregated network, corresponding to the union of the three biological networks used in this study, or using the multiplex-normalized density, proposed in the main manuscript. Table S4: Number of genes and number and percentage of differentially expressed genes (DEGs) retrieved in the active modules by the different methods in 30 runs. Table S5: Number of genes and number and percentage of differentially expressed genes (DEGs) retrieved in the active modules by the different methods in 30 runs. Only the statistics of the subnetworks with at least 15 nodes are reported here.

(PDF)