RNA sequencing (RNA-seq) is a powerful tool in biology that helps researchers understand how genes are expressed under different conditions.
However, current methods for detecting differentially expressed (DE) genes can be limited by overdispersion, where the variance in fragment counts is larger than expected, especially with small sample sizes. To address this, researchers at Penn State University developed a new approach called DEHOGT (heterogeneous overdispersion genes testing). DEHOGT integrates information from all samples to better model overdispersion across conditions, improving the detection power of DE genes even with limited replicates.
Illustration of the distribution density functions by fitting the read counts from microglia cells with 1) Poisson distribution, 2) negative binomial distribution, and 3) quasi-Poisson distribution, respectively. Compared with the Poisson distribution, both the negative binomial and quasi-Poisson distributions provide better approximation by capturing the overdispersion in read counts.
In tests using synthetic RNA-seq data, DEHOGT outperformed popular methods like DESeq2 and EdgeR by identifying more DE genes, particularly those relevant to microglial cells under different stress hormone treatments. This method promises to enhance our understanding of gene expression dynamics and biological responses, offering insights that could advance future research and therapeutic strategies.
Availability – The experiment description and algorithm implementation are available via the following weblinks: https://github.com/xiaobai0518/DEHOGT.
Yuan Y, Xu Q, Wani A, Dahrendorff J, Wang C, Shen A, et al. (2024) Differentially expressed heterogeneous overdispersion genes testing for count data. PLoS ONE 19(7): e0300565. [article]