Researchers often study how genes are turned on and off in single cells to understand what makes cells different from one another. One powerful way to do this is through RNA sequencing, which tells us how much RNA, a molecule made from genes, is present in a cell. But when scientists look at RNA from each of the two gene copies that every cell has, the results can be noisy, especially when only a few RNA molecules are captured. This makes it hard to detect subtle differences between the two copies, known as allelic imbalance.

To overcome this challenge, a team led by researchers at the Victor Chang Cardiac Research Institute, Australia has developed a new method called ASPEN. ASPEN uses smart statistical modeling to better estimate both the average activity and variability of each gene’s two copies within single cells. The method includes advanced steps to clean up how RNA reads are mapped and to carefully model how much variation is expected by chance.

Overview of ASPEN

LHS: Raw sequencing reads are mapped to a combined genome created by concatenating the two parental genomes. The C57BL/6J (GRCm38) genome serves as the reference, while strain-specific genomes are produced by integrating variant information (SNPs and indels) into the Bl6 genome. RHS: ASPEN employs the Bayesian shrinkage method to stabilize dispersion estimates across genes with varying expression levels. Using weighted log-likelihood, original dispersion estimates are shrunk toward the common trend dispersion — the expected level of variation for genes exhibiting similar expression levels. The posterior (shrunken) dispersion estimates facilitate the detection of state-specific and differential changes in the allelic ratio distribution.

When tested on both simulated data and real cells, ASPEN was about 30% better at detecting allelic imbalance than existing techniques. The researchers used it on mouse brain organoids, which are tiny lab-grown models of the brain, and on immune T cells. They found genes that did not fully turn off one gene copy when they should have, plus genes where only one copy was used randomly. They also noticed that genes essential for basic cell functions showed low variability between copies, suggesting strong control. In contrast, genes involved in brain development and immune function showed higher variability, pointing to flexible regulation.

Overall, ASPEN is a valuable new tool for researchers who want to understand how gene regulation varies from cell to cell, especially when looking at the activity of each gene copy.

Availability – ASPEN is available as an R package at https://github.com/ewonglab/ASPEN