A0991
Title: Methods for robust multi-study genomic data integration: Applications in infectious diseases research
Authors: Evan Johnson - Rutgers University (United States) [presenting]
Abstract: Despite widespread efforts to study the etiology of many prevalent infectious diseases, there exists broad heterogeneity in disease etiology driven by the diverse elements of host immune function and the myriad of potential disease-causing organisms. Several limiting constraints complicate the ability to use these existing data. One important challenge is the lack of molecular datasets with a large enough sample size to explore disease relationships and adequately train and validate new biomarkers. Another challenge is the lack of sufficient computational tools and platforms for data analysis and biomarker generation across multiple studies and cohorts. Resources for infectious disease biomarker generation and validation and data integration are discussed. This includes curated data platforms that provide existing gene expression data from public repositories with user interfaces to enable the interactive analysis of specific infectious diseases. Multi-study analytics and visualization are discussed, focused on ensembles of new and existing ID-related immune signatures. Novel machine learning methods are proposed for multi-study learning and unify the proposed resources to drive next-generation multi-study biomarkers for multiple infectious diseases. These data science tools and resources will accelerate research, leading to better mechanistic understanding and new biomarkers for infectious disease progression and heterogeneity.
