Color-enhanced image showing single-cell fluorescence measurements of protein and mRNA in the human lymph node. He et al. resolve RNA and protein localization at the subcellular level using multiple cycles of nucleic acid hybridization of fluorescent molecular barcodes.

See He et al.

Artistic rendering of a system for protein structure prediction. Chowdhury et al. present a deep learning method to predict a protein’s structure from its sequence alone, with applications to orphan and de novo–designed proteins.

See Chowdhury et al.

A model of 2&#x2032-O-hexadecyl (C16)-conjugated siRNA (orange) in complex with RISC (white) and mRNA (green). Brown et al. show that C16-conjugated siRNA achieves efficient, durable targeted gene silencing in the central nervous system, eye and lung.

See Brown et al.

Mt. Everest’s East Rongbuk Glacier was one of the glaciers sampled to create the comprehensive Tibetan Glacier Genome and Gene (TG2G) catalog, an inventory of microbial genetic diversity in this region.

See Liu et al.

Artistic impression of the Deep Visual Proteomics method developed by Mund et al., which is based on a combination of imaging, deep learning and mass spectrometry. Single cells are segmented, cell-typed, excised from tissue and analyzed for their protein content.

See Mund et al.

Micrograph of Cytaeis uchidae. A fluorescent protein expressed by this hydrozoan forms the basis of StayGold, a fluorescent protein with substantially increased photostability engineered by Hirano et al.

See Hirano et al.

An illustration of spirulina powder, which can be engineered to produce therapeutic proteins for human consumption. The method, developed by Jester et al., overcomes limitations in current genetic tools to generate stable spirulina lines expressing high levels of bioactive proteins at large scale.

See Jester et al.

Kuchroo et al. use data geometry and topology to develop Multiscale PHATE, a visualization tool for single-cell data. When applied to immune cells of COVID-19 patients, it helps find cell subgroups predictive of mortality.

See Kuchroo et al.

Artist’s impression of a phylogenetic tree. Konno et al. present a highly efficient distributed computing method for the reconstruction of evolutionary trees from very large datasets.

See Konno et al.

A synthetic biology process converts the one-carbon molecules CO and CO2 from steel-mill waste gases into the useful three-carbon molecules acetone and isopropanol. The method, demonstrated by Liew et al. at pilot industrial scale, provides a carbon-negative route for manufacturing the two compounds.

See Liew et al.

Rendering of the predicted structure of an RNA editor based on the compact Cas13bt enzyme presented by Kannan et al. Cas13bt is small enough to be delivered using adeno-associated viruses, even when additional functional domains, such as deaminases, are fused to the enzyme.

See Kannan et al.

Lotfollahi et al. present scArches, a method to contextualize new single-cell datasets within existing reference atlases without the need to share raw data.

See Lotfollahi et al.