Isoform diversity as a complementary dimension of gene expression

Most human genes produce multiple mRNA isoforms through alternative splicing and alternative transcription start and end sites. Isoforms of the same gene can differ in function, and their relative abundance varies between tissues and conditions. However, gene expression analysis is often restricted to the gene level, masking its diversification into mRNA and protein isoforms with potentially distinct functionality. Here we show, using an adaptation of Shannon evenness applied to 4,814 GTEx samples from 25 tissues, that isoform diversity constitutes a distinct dimension of gene expression, complementary to gene-level abundance and whole transcriptome diversity. Isoform diversity separated tissues based on non-differentially expressed genes and changed with age in a gene- and tissue-specific manner, most strongly in the brain and often without changes in overall gene expression levels. During neuronal differentiation, it increased most notably in RNA processing and splicing genes. Our results suggest that isoform diversity represents an independent dimension of gene expression and may help explain functional differences between tissues and cell states that are not captured by analyzing expression levels alone.