# ASPEN: Robust detection of allelic dynamics in single cell RNA-seq

## TL;DR

ASPEN is a new method that improves detection of allelic expression patterns in single-cell RNA-seq data, revealing regulatory dynamics in brain and immune cells.

## Contribution

ASPEN introduces a moderated beta-binomial model with adaptive shrinkage for robust allelic imbalance and variance detection in single-cell RNA-seq.

## Key findings

- ASPEN improves allelic imbalance detection sensitivity by ~30% compared to existing methods.
- ASPEN identifies genes with random monoallelic expression and incomplete X inactivation.
- Essential genes show reduced allelic variance, while neurodevelopmental and immune genes show increased variance.

## Abstract

Single-cell RNA-seq data from F1 hybrids provide a unique framework for dissecting complex regulatory mechanisms, but allelic measurements are limited by technical noise due to low counts. Here, we present ASPEN, a statistical method for modeling allelic mean and variance in single-cell transcriptomic data. ASPEN combines a sensitive mapping pipeline  with a moderated beta-binomial model and adaptive shrinkage to distinguish allelic imbalance and changes to allelic variance in single cells. In both simulated and empirical datasets, ASPEN achieves a ~30% increase in sensitivity over existing approaches for single-cell allelic imbalance detection. Applied to mouse brain organoids and T cells, ASPEN identifies genes with incomplete X inactivation, random monoallelic expression, and significant deviations in allelic variance. These results reveal reduced variance in essential genes, consistent with tight regulatory control, and increased variance at neurodevelopmental and immune loci, indicative of regulatory flexibility.

Hybrid organisms inherit two distinct copies (alleles) of each gene, one from each parent. By measuring how much each allele is expressed in individual cells, we can detect regulatory differences that are invisible in bulk tissue samples. However, single-cell measurements are sparse, making it difficult to separate true biological signals from technical noise. We developed ASPEN, a statistical framework that stabilizes these noisy measurements to reliably estimate allelic expression. ASPEN improves the detection of allelic imbalance by up to 30% compared with existing methods and, uniquely, quantifies allelic variance, how much allele usage fluctuates across cells. Applying ASPEN to developing brain cells and immune cells, we discovered that essential “housekeeping” genes maintain remarkably stable allelic ratios, while genes involved in brain development and immune function show much greater variability. We also identified genes with random monoallelic expression and incomplete X chromosome inactivation. By capturing these diverse patterns, ASPEN provides a clearer view of the dynamic regulatory mechanisms that shape gene expression.