Strategies for molecular classification of asthma using bipartite network analysis of cytokine expression

Abstract

Asthma is a chronic inflammatory disease of the airways that leads to various degrees of recurrent respiratory symptoms affecting patients globally. Specific subgroups of asthma patients have severe disease leading to increased healthcare costs and socioeconomic burden. Despite the overwhelming prevalence of the asthma, there are limitations in predicting response to therapy and identifying patients who are at increased risk of morbidity. This syndrome presents with common clinical signs and symptoms; however, awareness of subgroups of asthma patients with distinct characteristics has surfaced in recent years. Investigators attempt to describe the phenotypes of asthma to ultimately assist with diagnostic and therapeutic applications. Approaches to asthma phenotyping are multifold; however, it can be partitioned into 2 essential groups, clinical phenotyping and molecular phenotyping. Innovative techniques such as bipartite network analysis and visual analytics introduce a new dimension of data analysis to identify underlying mechanistic pathways.

Fig. 1

Network basics. A simple network is composed of nodes and edges. Here the network attempts to describe the relationship between the drug and clinical reaction. Red nodes denote the drug. The white nodes denote the reaction. The edges connect the nodes. White nodes (reaction) are distributed in space according to their degree of association to the red nodes (chemical agent). By this representation subsets of reactions and drugs can be identified. For example, it appears that DDT, Hepatachor, and Dieldrin can lead to similar side effects

Fig. 2

Network analysis of asthma patients and cytokines. The network reveals cytokine and patient clusters