RNA-seq analysis of LNCaP data under androgen and vitamin D treatments

Androgens and vitamin D are two essential elements in the study of prostate cancer. Investigations on these two hormones have been conducted for decades, but the mechanisms of their interplay are not yet elucidated. In this study, RNA-seq data of the LNCaP cell line based on a factorial design of these two supplements are analysed. DE analysis was conducted using three popular pipelines, DESeq2, edgeR and limma+voom. The results showed that more DE genes were called by DESeq2 than by the other pipelines. A microarray dataset of a similar factorial design and ChIP-Atlas datasets on AR and VDR were used to assess the pipelines. DESeq2 showed higher consistency with both sources with respect to the DE results under vitamin D treatment. The DE results given by DESeq2 revealed several vitamin D-induced genes, which had oncogenic roles. That indicated the paradoxical role of vitamin D in prostate cancer. The GSEA results revealed that the crosstalk between androgen and vitamin D could enhance the function of vitamin D in cell-cycle arrest. Apart from data analysis, a systematic review of benchmark papers that assessed normalisation performance was produced. This review provides a reservoir of information for researchers who seek to find a proper normalisation method or build their own assessment experiments. This review offers a guide on how a tool for RNA- seq data should be assessed. Furthermore, a new normalisation method, termed MODE, based on the mode of the log-ratio between two samples, was proposed. Both simulated and real data revealed that MODE should be an optimal alternative to the popular DESeq RLE and edgeR TMM. However, applying MODE to the androgen/vitamin D data did not show much difference, compared to the standard DESeq2 pipeline.