Bulk RNA sequencing remains one of the most widely used methods for measuring gene activity because it is cost effective and works well for large research studies. However, it averages signals from millions of cells, making it difficult to determine which specific cell types are responsible for the observed biological changes.

Researchers from the University of Innsbruck have developed Rectangle, a new computational framework that combines the strengths of bulk RNA sequencing with information from single-cell RNA sequencing. The software helps researchers estimate the abundance of individual cell types and cell states within complex tissue samples, providing much greater biological detail without requiring every sample to undergo expensive single-cell sequencing.

Rectangle performs robust multiscale deconvolution informed by single-cell transcriptomics

Rectangle takes annotated scRNA-seq data and a bulk RNA-seq mixture as input. In the signature-building phase, it performs bootstrapped pseudobulking of the scRNA-seq data, followed by signature-gene selection based on log-fold-change (logFC) and p-value, and signature-matrix optimization (minimizing the condition number and correcting for cell-type-specific mRNA content bias). Based on this, it constructs two signature matrices: a “direct signature”, resolving individual cell types, and a “clustered signature” grouping transcriptionally similar cell types. In the deconvolution phase, Rectangle first uses the clustered signature to deconvolve the bulk data into coarse cell-type estimates. These are then used to constrain a second deconvolution step that leverages the direct signature to resolve individual cell-type fractions. Finally, the resulting estimates are scaled to account for unknown cellular content not represented in the reference.

Many existing deconvolution tools use single-cell RNA sequencing data as a reference to estimate the cellular composition of bulk samples. While these approaches have been valuable, they often struggle to distinguish closely related cell populations, become slow when analyzing very large single-cell reference datasets, or fail when tissues contain cell types that are missing from the reference data.

Rectangle was designed to overcome these limitations. It performs multiscale deconvolution, allowing researchers to examine tissues at several levels of cellular detail. Rather than limiting analysis to broad cell categories, the framework can identify increasingly refined cell populations and functional cell states as the available reference data become more detailed.

Another important feature is Rectangle’s ability to account for unknown cellular content. Real biological samples frequently contain rare or previously uncharacterized cell types that are not represented in existing single-cell reference datasets. Instead of forcing these cells into incorrect categories, Rectangle explicitly models unknown cellular contributions, improving the accuracy of the final results.

The researchers evaluated Rectangle against multiple existing deconvolution methods using a wide range of benchmark datasets. Across nearly every performance measure, including accuracy, scalability, computational efficiency, and resistance to signal spillover between related cell types, Rectangle consistently performed among the strongest methods.

Because bulk RNA sequencing is significantly less expensive than single-cell RNA sequencing, many large clinical and population studies rely on bulk datasets. Rectangle makes it possible to extract much richer cellular information from these existing datasets while maintaining the scalability needed for studies involving thousands of samples.

As RNA sequencing continues to generate increasingly large collections of transcriptomic data, computational tools like Rectangle help bridge the gap between affordable bulk RNA sequencing and the detailed cellular insights provided by single-cell technologies. This approach could accelerate biomarker discovery, improve studies of disease progression, and enable researchers to monitor changes in cell populations across large patient cohorts.

Availability – Rectangle is available as an open-source Python package (https://github.com/ComputationalBiomedicineGroup/Rectangle).