BLASE: bulk linkage analysis for single cell experiments - teasing out the secrets of bulk transcriptomics with...

Transcriptomics has profoundly improved our knowledge of cells. The advent of single-cell transcriptomics has enabled researchers to investigate the changes which individual cells undergo, such as cell type differentiation. Bulk RNA-seq is more practical in both cost and ease, but cannot elucidate cell type specific trajectories. Deconvolution methods can estimate cell types in RNA-seq data, but there is a need for methods characterising their position on a trajectory.

We present a new method called BLASE, which discretises the pseudotime of a scRNA-seq trajectory, and then uses Spearman correlation to infer the closest matching period of pseudotime to a bulk RNA-seq sample. Bootstrapping provides confidence intervals around the correlation, and subsequently informs "strong" calls. BLASE can discretise pseudotime using several different methods, provides heuristics for hyperparameter selection, and enables the visualisation of results.

BLASE performs correctly and outperforms other tools. In simulated scRNA-seq data it accurately identified the correct pseudotime bin for 10 pseudobulked pseudotime bins, whereas other tools we tested ranged in accuracy from around 40-90%. In experimental data, BLASE correctly mapped all pseudobulked pseudotime bins, compared to a range of around 20-75%. We tested BLASE on use cases with published single-cell and bulk transcriptomics. On 10x Visium spatial data, BLASE could resolve the spatio-temporal process of keratinocyte differentiation. When mapping a time-course of 48 hour microarray data to the Plasmodium falciparum lifecycle, BLASE correctly identified the predominant cell type of synchronised cells in 48/48 samples (30 strong). Finally, BLASE identified a developmental rate difference in P. falciparum grown with or without heat shock conditions. Of 345 genes differentially expressed, 142 were attributed to developmental rate differences. The remaining 203 should represent the true signal better, and revealed new gene ontology terms.

BLASE outperforms existing tools in simulated and experimental data. It can be used to a) annotate scRNA-seq data from existing RNA-seq, b) identify progress of RNA-seq data through a process captured in scRNA-seq, and c) be used to correct developmental differences in differential expression analysis. BLASE is released as an open-source R package under the GPL3 license, and is available on Bioconductor.

Not applicable.