Analyze a Human Disease by Network Modeling

Although there is a continuous generation of molecular ‘omics’ data from biomaterials derived from patients, there is a major challenge on how to analyze and integrate these data in order to identify the central regulators of the pathogenesis of a human disease.

Network analysis could contribute to the construction of a human disease molecular Interactome. Network theory is a branch of applied mathematics that uses the concepts of graph theory. The development of network theory was led by applications to real-world examples, such as social networks and technological (Internet) networks. The structure of complex molecular interactions in a human disease can be represented by networks and not by linear signaling pathways, a methodology followed by most researchers today. A network could reveal the positive and negative feedback loops and information exchange between the different signaling pathways. Overlooking the loop circuits led to several failures of drugs developed by pharmaceutical companies.

In a human disease network, nodes could correspond to genes, proteins, or metabolites, whereas the edges will represent the interactions, causal influences, or correlations between them. To detect the central molecular regulators in the pathogenesis of a human disease, the disease network could be compared to random graphs with defined statistical properties. We can build networks based on the specific characteristics described above. For example, we can build a disease metabolic network of the metabolites deregulated in patients and the chemical reactions that connect these metabolites. A disease transcriptional network can be constructed by identifying transcriptional interactions between deregulated coding and noncoding RNAs. A protein network will include the deregulated protein–protein interactions in specific cell types. The next step would be to integrate all these networks aiming to construct the human disease molecular interactome.

An important question is how practically could we construct and visualize these disease molecular networks?

The first step would be the generation of the ‘omics’ data from patient biomaterials. Next, the dynamics and regulatory patterns of the potential gene, protein and metabolite interactions should be described by a mathematical graph. The graph could be constructed using Boolean network analysis, Bayesian network analysis, ordinary and partial differential equation systems, or stochastic processes. A graph consists of a discrete set of nodes (N) and edges (E), which are defined as pairs of nodes. The nodes could be genes, proteins, or metabolites deregulated in a human disease, and the edges will show direct or indirect interactions between the nodes. Each network will be characterized by its statistical properties. The node that shares an edge with another node is called neighbor, and the number of neighbors is called the degree or node size (k). A distinction between in-degrees (kin) and out-degrees (kout) refers to incoming and outgoing edges, respectively. The gene, protein, or metabolite with the high number of incoming and outgoing edges is called central regulator or hub. The hubs identified by network analysis could be used as disease drug targets because they will be central regulators of the disease networks.

In the last few years, multiple software programs have been developed for systems biology purposes. A visit to the software guide at website will reveal more than 100 different software developed for network and modeling analyses. JSIM is a Java-based software able to build quantitative networks and can be used from a web browser. In addition, the CellDesigner is another Java-based tool that could show visually appealing graphical representations of the disease networks . Furthermore, E-Cell is a Python-based software able to model, simulate, and analyze large-scale networks and systems. These tools require high knowledge of mathematics and computer programming, revealing the importance of integrating computational biologists in biomedical research. In addition to computational tools, there are commercially available tools that do not require computing knowledge. A user-friendly software, called Ingenuity Pathway Analysis (IPA), constructs molecular networks based on experimental and literature-based data. These computational tools can integrate molecular ‘omics’ data into networks for each of the cellular populations involved in the pathogenesis of any human disease.