Global prediction of chromatin accessibility using small-cell-number and single-cell RNA-seq

Abstract

Conventional high-throughput genomic technologies for mapping regulatory element activities in bulk samples such as ChIP-seq, DNase-seq and FAIRE-seq cannot analyze samples with small numbers of cells. The recently developed low-input and single-cell regulome mapping technologies such as ATAC-seq and single-cell ATAC-seq (scATAC-seq) allow analyses of small-cell-number and single-cell samples, but their signals remain highly discrete or noisy. Compared to these regulome mapping technologies, transcriptome profiling by RNA-seq is more widely used. Transcriptome data in single-cell and small-cell-number samples are more continuous and often less noisy. Here, we show that one can globally predict chromatin accessibility and infer regulatory element activities using RNA-seq. Genome-wide chromatin accessibility predicted by RNA-seq from 30 cells can offer better accuracy than ATAC-seq from 500 cells. Predictions based on single-cell RNA-seq (scRNA-seq) can more accurately reconstruct bulk chromatin accessibility than using scATAC-seq. Integrating ATAC-seq with predictions from RNA-seq increases the power and value of both methods. Thus, transcriptome-based prediction provides a new tool for decoding gene regulatory circuitry in samples with limited cell numbers.

BABEL enables cross-modality translation between multiomic profiles at single-cell resolution

“…For benchmarking the opposite RNA to ATAC translation, we are not aware of any prior methods that perform single-cell prediction. BIRD is a recently developed relevant method, although it is trained to make cluster-aggregated predictions of ATAC signals instead ( 32 , 33 ), and consequently may be less flexible when applied to cell types it has not seen before. BABEL compares favorably to BIRD as well ( SI Appendix , Fig.…”

A systematic evaluation of single-cell RNA-sequencing imputation methods

“…To rank methods across datasets, we first averaged the median SCCs across datasets within the same experimental protocol (e.g., UMI-based or plate-based) and then averaged these averages across protocol. In the analyses above, pseudobulk was used as a reference for the approximate upper bound of the single-cell imputation performance since both bulk and pseudobulk profiles try to measure a cell population's average behavior, and the correlation between a pseudobulk and the corresponding bulk profile is expected to increase as one pools an increasing number of cells to create the pseudobulk sample [74,75]. Note that unlike single-cell profiles, pseudobulk cannot characterize cell-to-cell variability.…”

Single-nucleus RNA and ATAC sequencing reveals the impact of chromatin accessibility on gene expression in Arabidopsis roots at the single-cell level

“…Ultimately, each plant cell will activate or repress specific sets of genes to fulfill the biological functions inherent to their cell type and their response to environmental signals. In animal science, single-cell RNA sequencing (scRNA-seq) and single-cell ATAC-seq (Pott and Lieb, 2015) technologies have been successfully applied across various cell types and tissues to better understand the impact of the dynamic accessibility of chromatin on gene expression (Buenrostro et al, 2015(Buenrostro et al, , 2018Norrie et al, 2019;Zhou et al, 2019) Recently, scRNA-seq approaches have been applied to Arabidopsis root protoplasts, allowing the accurate characterization of the transcriptional profiles of thousands of cells and their differential regulation in mutants or in response to a stress (Denyer et al, 2019;Jean-Baptiste et al, 2019;Ryu et al, 2019;Shulse et al, 2019;Zhang et al, 2019). These studies revealed the power of single-cell technologies to establish the transcriptomic maps of various Arabidopsis root cells and cell types and the dynamic regulation of gene expression during cell development.…”