A high-throughput system for using single-cell CRISPR screens to study the genetic determinants of chromatin accessibility is described in Nature Biotechnology this week. Combining CRISPR screens with single-cell RNA sequencing has enabled researchers to connect genetic perturbations with changes in gene expression and phenotypes across the transcriptome, yet methods to examine changes in the epigenome have limited throughput. Aiming to study how genetic perturbations affect chromatin states, scientists from the New York Genome Center developed a platform for scalable, pooled CRISPR screens with single-cell assay for transposase-accessible chromatin using sequencing profiles. They use their approach, dubbed CRISPR-sciATAC, to target 105 chromatin-related genes, generating chromatin accessibility data for roughly 30,000 single cells, and correlate the loss of specific chromatin remodelers with changes in accessibility globally and at the binding sites of individual transcription factor. "CRISPR-sciATAC can be applied to the study of diverse phenotypes and diseases, and to understand the interaction between genetic changes and genome-wide chromatin accessibility," they write.
A University of Paris team reports a novel statistical method for extracting gene signatures at the single-cell level from sequencing data in Nature Biotechnology. While high-throughput single-cell sequencing is widely used for cellular characterization, current methods for exploring cell-type diversity are limited by the high dimensionality and high levels of technical and biological noise associated with single-cell measurements. To overcome this issue, the investigators developed Cell-ID, a multivariate approach for extracting a gene signature for each individual cell in a study. They demonstrate the tool on data from multiple human and mouse samples, as well as comprehensive mouse cell atlas datasets, and show that Cell-ID signatures are reproducible across different donors, tissues of origin, species, and single-cell omics technologies. It can also be used automatic cell-type annotation and cell matching across datasets, they write.
Finally, a new normalization procedure for single-cell RNA sequencing data is presented in this week's Nature Biotechnology by a trio of University of Basel scientists. "Using only minimal assumptions, we derive from first principles a Bayesian method that corrects not only for the finite sampling associated with the capture and sequencing of mRNAs but also for the Poisson noise inherent in the gene expression process itself," they write. The method — called sampling-noise-corrected inference of transcription activity, or Sanity — estimates expression values and associated error bars directly from raw unique molecular identifier counts without any tunable parameters. In testing with both simulated and real-world datasets, Sanity outperformed existing methods on basic downstream processing tasks. The team also demonstrates that the other methods produce a representation of the data that is distorted in one or more respects.