Research Briefing in 2023

Using an experimental and computational framework inspired by compressed sensing, we greatly reduced the number of measurements needed to run Perturb-seq. Our compressed Perturb-seq strategy relies on collecting measurements comprising random linear combinations of genetic perturbations, followed by deconvolving the perturbation effects on the transcriptome using sparsity-exploiting algorithms.

We designed a method for fast aptamer selection by integrating biomaterials science, engineering principles and biology. Aptamer candidates dynamically interacting with immobilized targets in a three-dimensional, non-fouling and macroporous polyethylene glycol hydrogel were rapidly enriched and selected with high affinity against five protein targets.

Most features of a cell are determined by gene programs — sets of co-expressed genes that execute a specific function. By incorporating existing knowledge about gene programs and cell types, the Spectra factor analysis method improves how we decode single-cell transcriptomic data and offers insights into challenging tumor immune contexts.

The detection of mobile genetic elements is crucial for exploring the ecology and evolution of microbial communities, and it has diverse implications in biotechnology and public health. geNomad is a computational framework that enables researchers to precisely identifiy and annotate plasmids and viruses in sequencing data on a large scale.

By using fixed charges to engineer a strong electroosmotic flow, we achieve the unidirectional transport of natural polypeptides across nanopores. Our approach enables native proteins to be transported enzymatically and non-enzymatically in the absence of denaturant and electrophoretic tags, with potential applications for protein sequencing.

BugSigDB is a community-editable wiki that harmonizes how key microbial differential abundance methods and results are reported, identifying rare and common patterns across the literature of published host-associated microbiome studies.

Data integration between weakly linked single-cell modalities is challenging using existing methods. Therefore, we developed MaxFuse to enable matching and integration between cells from modalities such as single-cell spatial proteomic datasets and single-cell transcriptomic datasets, or other modalities where features are only weakly correlated.

Although base and prime editors can be highly efficient in human hematopoietic stem cells, we find they can cause adverse cellular responses, including reduced engraftment and the generation of DNA double-strand breaks and genotoxic byproducts, albeit at a lower frequency than Cas9. We also find that base editors increase the genome-wide mutagenic load.

GEARS, a machine learning model informed by biological knowledge of gene–gene relationships, effectively predicts transcriptional responses to multi-gene perturbations. GEARS can predict the effects of perturbing previously unperturbed genes and detects non-additive interactions, such as synergy, when predicting combinatorial perturbation outcomes. Thus, GEARS expands insights gained from perturbational screens.

Directly analyzing the role of the gut microbiota during the course of infection with human-specific pathogens is not possible with existing methods. To overcome this problem, we developed a germ-free precision mouse model, which we used to compare the acquisition, replication and pathogenesis of HIV and Epstein–Barr virus.

A systematic examination of eight different single-cell assay for transposase-accessible chromatin by sequencing (scATAC-seq) technologies revealed marked differences in the complexity of sequencing libraries and the specificity of DNA tagmentation that they achieve. Our pipeline for universal mapping of scATAC-seq data (PUMATAC) allowed a fair benchmarking of existing methods and enables the seamless integration of future datasets and technologies.

Wild three-dimensional imaging of solvent-cleared organs (wildDISCO) uses conventional antibodies for whole-body staining in mice, creating comprehensive biological atlases of nerves, blood vessels and lymphatics. It uncovers pathological changes, such as tertiary lymphoid structures in cancer, and it enables the precise tracking of therapeutic molecules and cells, enhancing our understanding of disease pathology and treatment.

We present an algorithm, SComatic, that can be used to directly detect somatic mutations in single-cell data sets without using a reference sample. This method opens the possibility of studying clonal relationships among cells, mutational processes at single-cell resolution, and the impact of somatic mutations on cell function in development and disease.

TnpB proteins are hypercompact RNA-guided DNA endonucleases. By systematically mining and characterizing TnpB proteins from the IS605 family, we identified potent genome editors and established a framework for high-throughput annotation and screening of TnpB systems in prokaryotic genomes.

Generating A-to-C transversions in specific targets via base editing technology has been challenging. By fusing an evolved alkyladenine DNA glycosylase with an engineered adenine deaminase TadA-8e variant and nickase Cas9, we have developed A-to-C base editors that generate precise and efficient A-to-C transversions in cells and in mouse embryos, expanding the possible applications of base editing.

We created miniature and flexible polymer fibers equipped with optoelectronic microdevices, microfluidics and electrodes that can be implanted in anatomically disparate organs such as the brain and the intestine. The microelectronic fibers can be operated wirelessly to sense and manipulate brain and gut neural circuits in untethered, behaving mice.

Existing methods to infer cell–cell communication from single-cell RNA-sequencing data fail to leverage the full information structure of the data, generally by operating at the level of the cell type or cluster. We describe a framework called Scriabin to perform this analysis at the level of the individual cell.

The fusion of a programmable transcription-activator-like effector (TALE) protein with a nickase, in conjunction with a deaminase, enables efficient and strand-selective DNA base editing. This approach has the potential to advance our understanding and treatment of diseases associated with mutations in the mitochondrial or nuclear genome.

Minigraph-Cactus, a method to efficiently combine multiple reference genome assemblies into a pangenome reference graph, can be used to improve accuracy of read mapping and variant calling compared with a single reference in downstream applications.

By linking transgene expression with that of a housekeeping gene, SLEEK (selection by essential-gene exon knock-in) enables efficient knock-in of complex cargos in a variety of clinically relevant cell types. Using SLEEK, we were able to substantially improve the tumor suppression ability and in vivo persistence of a cell therapy.