Question

In RNA-Seq analysis, why must gene expression counts be normalized by both library size and gene length before comparing expression levels between samples?

Accepted Answer

RNA-Seq counts represent the number of sequencing reads that align to a specific gene, but these raw values are influenced by two technical biases that prevent direct comparison between samples. First, library size bias occurs because different sequencing runs produce varying totals of reads. If one sample is sequenced more deeply than another, it will have higher counts for every gene simply due to the total number of reads generated, rather than actual biological differences. Normalizing for library size adjusts these counts so that samples can be compared on an equal scale regardless of how many total reads were produced. Second, gene length bias occurs because longer genes are more likely to be fragmented into more pieces and thus produce more reads than shorter genes, even if both genes are expressed at the same biological level. A longer gene will naturally capture a larger portion of the sequencing output than a shorter gene, creating the false appearance of higher expression. Normalizing for gene length ensures that the expression value reflects the concentration of RNA molecules rather than the physical length of the gene transcript. By correcting for both total sequencing depth and gene transcript length, these normalizations allow researchers to compare the relative abundance of RNA molecules across different genes within a sample and across different samples in an experiment.

Home → All Courses → Science and Mathematics Courses → Computational Biology → Flashcard

In RNA-Seq analysis, why must gene expression counts be normalized by both library size and gene length before comparing expression levels between samples?