Testing for differentially expressed genetic pathways with single-subject N-of-1 data in the presence of inter-gene correlation

What is it about?

Interpreting RNA measurements from cancer patients is a difficult task due highly individualized disease characteristics yet offers an opportunity to provide personalized medicine. The statistical methodology developed in our work builds on existing methods that identify sets of genes (pathways) that are dysregulated in the cancer tissue, within each patient. The important contribution of the new method lies in an adjustment for correlation in gene-level quantification, which has been ignored in previous works (and, when ignored, can drastically inflate errors in statistical testing).

Featured Image

Why is it important?

The new method, the Clustered-T, provides a clinical tool to help physicians and medical researchers treat patients based on individual profiles of molecular dysfunction. This has the potential to support decision making despite the massive complexity and huge data inherent to the genomic era. Further, the experimental design is economically efficient in that only two samples are needed from the patient to promote translation into the medical system.