FET: III: Small: Innovative Approaches for Bias Correction and Systems-level Analysis in Integrated Multi-omics Data

Identifying impacted pathways and changes in biological processes is important because it provides insights into the biology underlying conditions beyond the detection of differentially expressed genes. Because of the importance of such analysis, more than 100 pathway-analysis methods have been developed thus far. However, as all these methods are biased toward well-studied conditions, such as cancer diseases, the accuracy of pathway-analysis methods is severely compromised, especially when investigating non-cancer diseases and phenotypes.

More importantly, existing methods are limited to the analysis of a single cohort or data type, making them sensitive to biological heterogeneity and unable to analyze complex diseases that involve multiple molecular levels. This project aims to bridge these gaps by designing an analysis pipeline that will tackle bias correction and data integration in one computational framework.

This will have a great impact in many research and public health areas by facilitating the identification of putative molecular causes of disease, as well as the identification of potential therapeutic interventions and their possible side effects. This project also includes a systematic outreach program involving Primarily Undergraduate Institutions (PUI) across Nevada, especially minority- and Hispanic-serving institutions: College of Southern Nevada and Nevada State College.

The applications presented here will enable students to conduct exciting scientific research without the need to perform wet-lab experiments. Other planned outreach activities involve summer workshops for K-12 local schools from Washoe County School District.

Most pathway-annotation databases have important limitations. Some of these limitations are related to the domain (e.g., focused on cancer alone), others to the types of data included (e.g., only expression data), and still others to the level of detail chosen to describe the phenomenon. Further, pathway-analysis methods are subject to systematic bias due to unrealistic assumptions and overfitting.

Another pain point is the inability to easily include multiple cohorts and multiple types of -omics data in the same analysis. This project will provide a framework that allows researchers to retain their preferred pathway methods while correcting for method bias and integrating multiple data types and datasets. The goal of this project will be achieved by two thrusts: 1) design a methodology for bias correction and consensus analysis of pathway methods, and 2) develop a novel approach for the flexible integration of multi-cohort and multi-omics data.

The framework will be designed so that it can be applied in conjunction with any existing pathway-analysis method to correct for bias, incorporate knowledge from different databases and integrate data of different types. The project also includes a systematic evaluation plan of the proposed methodologies using more than 100 datasets with known mechanisms.

The research team will deliver an implementation that supports several widely used methods for many model organisms.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.