Fused inverse-normal (FIN) method for meta analysis

Fused inverse-normal method for meta-analysis of RNA-seq data.

This repository contains an implementation of fused inverse-normal (FIN) method for meta-analysis of multiple but related RNA sequencing (RNA-seq) studies. The current code demonstrates the method for combination of p-values from three different real glioblastoma RNA-seq studies and can be adapted to any number of RNA-seq gene expression studies.

Although designed for meta-analysis in the context of RNA-seq data, the method can be applied in other contexts when the variables of interest assume two distinct states across multiple studies.

Published article and citation

Prasad, B., Li, X. Fused inverse-normal method for integrated differential expression analysis of RNA-seq data. BMC Bioinformatics 23, 320 (2022). https://doi.org/10.1186/s12859-022-04859-9

Prerequisites:

• Results from per-study differential analysis for RNA-seq studies. Popular methods such as DESeq and edgeR can be used for this step. These results should at least contain the raw p-values and log_2(FC) (logFC) from the individual differential analaysis.
• Assessment of the underlying assumption that p-values for all genes obtained from per-study differential analysis are uniformly distributed under the null hypothesis. Usually this assumption is not satisfied in case of RNA-seq data but filtering of weakly expressed genes using the method described in Rau et al. (2014)[1] or using the counts per million criteria described in Raithel et al. (2016)[2] circumvents this difficulty to a significant extent.
• R version 3.6.0 or above.

References

1. Rau A, Marot G, Jaffrézic F. Differential meta-analysis of RNA-seq data from multiple studies. BMC bioinformatics. 2014 Dec 1;15(1):91. DOI
2. Raithel S, Johnson L, Galliart M, Brown S, Shelton J, Herndon N, Bello NM. Inferential considerations for low-count RNA-seq transcripts: a case study on the dominant prairie grass Andropogon gerardii. BMC genomics. 2016 Dec 1;17(1):140. DOI