Phantom: investigating heterogeneous gene sets in time-course data.

Motivation: Gene set analysis is a powerful tool to study the coordinative change of time-course data. However, most existing methods only model the overall change of a gene set, yet completely overlooked heterogeneous time-dependent changes within sub-sets of genes. Results: We have developed a novel statistical method, Phantom, to investigate gene set heterogeneity. Phantom employs the principle of multi-objective optimization to assess the heterogeneity inside a gene set, which also accounts for the temporal dependency in time-course data. Phantom improves the performance of gene set based methods to detect biological changes across time. Availability and implementation: Phantom webpage can be accessed at: http://www.baylorhealth.edu/Phantom. R package of Phantom is available at https://cran.r-project.org/web/packages/phantom/index.html. Contact: jinghua.gu@bswhealth.org Supplementary information: Supplementary data are available at Bioinformatics online.