Network Clustering

Network Clustering

This project is an objective machine learning (ML) approach to detect differential epigenetic-networks (DENs) through network clustering using DNAm data in a population before disease manifestation, e.g., at birth using childhood asthma as an exemplar.

It is increasingly recognized that asthma is a consequence of joint activities among genes rather than of individual genes’ independent contributions. Examining individual contributions of CpGs on asthma development is biologically ungrounded, and, more significantly, increases the risk of incomplete and/or mis-leading findings. Such limitations have been widely agreed and methods to address joint activities have been proposed including approaches to identify differentially methylated regions (DMRs) and those for detection of gene networks. However, the DMR approach is potentially flawed, since almost all DMRs are inferred based on effects of each individual CpG rather than joint effects of a set of CpGs. Definfing networks amongst genes or CpGs allow better understanding of their concerted effects. Current network constructions either focus on a population as a whole or on a specific group. Constructing one network for a general population lacks purity, since it overlooks underlying heterogeneity among subjects (e.g., heterogeneity in asthma risk). Whilst building a network for a group of disease patients will substantially limit the ability to predict and prevent disease, due to potential reverse-causation on gene activities. Currently, no methods are available that address heterogeneity while building networks.

• Health, Wellness and Insurance
• 97% Customer Satisfaction
• Family Scholarship Program