DOMINO: a novel algorithm for network-based identification of active modules with reduced rate of false calls

Abstract

Network-based module discovery (NBMD) methods are central to analysis of omics data. Such algorithms receive a gene network and nodes' activity scores as input and report sub-networks (modules) that are putatively biologically active. Although such methods exist for almost two decades, only a handful of studies attempted to compare the biological signals captured by different methods. Here, we systematically evaluated six popular NBMD methods on gene expression (GE) and GWAS data . Notably, we observed that GO terms enriched in modules detected by these methods on the real data were often also enriched after randomly permuting the input data. To tackle this bias, we designed a method that evaluates the empirical significance of GO terms reported as enriched in modules. We used the method to fashion five novel performance criteria for evlautating NBMD methods. Last, we developed a novel NBMD algorithm called DOMINO. In extensive testing on GE and GWAS data it outperformed the other six algorithms. Software is available at https://github.com/Shamir-Lab/.

Network-based approaches for modeling disease regulation and progression

“…They showed that many DNE methods mostly learn from the node degree distributions rather than from the encoded biological knowledge in the edges of PPI networks [159] . By assessing the gene ontology enrichment of DNE methods on randomly permuted input data, Levi et al questioned the context-specificity of the existing methods [24] . Since diseases with higher morbidity or mortality and their associated proteins are studied more extensively, PPI networks are subject to study bias [160] .…”

“…Links to code/tools are provided in Supplementary Table 1 . Tool Input data type (s) Algorithm Type Example application Reference SigMod GWAS Aggregate score Identify functionally and biologically relevant genes in childhood-onset asthma [18] [18] IODNE Gene expression Aggregate score Identify potentially novel target genes for drug selection in triple-negative breast cancer [19] [19] PCSF Multi-omics data (gene expression, mutation profiles, or copy number) Aggregate score Extract subnetworks of enriched metabolite interactions in multiple sclerosis [36] [20] Omics Integrator Gene expression Aggregate score Link α-synuclein to multiple parkinsonism genes and druggable targets [37] [21] MuST Disease-associated genes (derived from GWAS or DEG analyses) Aggregate score Investigation of coagulation pathway in COVID-19 [38] [22] ROBUST Disease-associated genes (derived from GWAS or DEG analyses) Aggregate score Identify an oxidative stress module in multiple sclerosis [23] [23] DOMINO Disease-associated genes (derived from GWAS or DEG analyses) Aggregate score Integrated as the downstream analysis step in a splicing-aware framework for time course data analysis [39] [24] KeyPathwayMiner Gene expression / multi-omics data Module cover Reveal epigenetic targets in SARS-CoV-2 infection, used together with gene co-expression networks [40] [25] ModuleDiscoverer Gene expression Module cover Identify regulatory modules o...…”

“…ROBUST is another method using MWST which offers improved robustness compared to MuST by enumerating pairwise diverse rather than pairwise non-identical disease modules [23] . Levi et al demonstrated that most DNE methods using gene expression data do not fully exploit the information contained in this data [24] . They tried to address this issue in DOMINO by aiming to find disjoint connected Steiner trees in which the active genes (seeds) are over-represented [24] .…”

“…Levi et al demonstrated that most DNE methods using gene expression data do not fully exploit the information contained in this data [24] . They tried to address this issue in DOMINO by aiming to find disjoint connected Steiner trees in which the active genes (seeds) are over-represented [24] .…”

Lipid network and moiety analysis for revealing enzymatic dysregulation and mechanistic alterations from lipidomics data

“…While this assumption is not always valid, it can provide hypotheses about potentially disturbed mechanisms for further validation experiments. Many algorithms have been developed over the years [3, 35, 14, 41, 20], mainly with a focus on protein-protein interaction (PPI) or gene-regulatory networks. A dedicated method for metabolomics data is included in the MetExplore analysis and visualization software [12].…”