Gene regulatory networks articles from across Nature Portfolio

Accuracy of gene regulatory network inference is increased by combining multiome single-cell and atlas-scale bulk data.

Hepatoblastoma (HB) is the most frequent paediatric liver tumour with heterogeneous cellular phenotypes that influence clinical outcomes. Here, the authors integrate bulk, single-cell, and spatial multi-omics to characterise HB cells, and find that clonal evolution and epigenetic plasticity shape response to therapy.

Characteristic genes or proteins driving continuous biological processes are difficult to uncover from noisy single-cell data. Here, authors present DELVE, an unsupervised feature selection method to identify core molecular features driving cell fate decisions.

Using patient-derived induced pluripotent stem cells, the authors show that the transcription factor GFI1B and the lysine demethylase KDM1A/LSD1 promote gene programs while repressing those involved in myeloid differentiation.

With biomedical sciences quickly outgrowing many other application areas in terms of data generation, there is a unique opportunity for life sciences to become one of the greatest beneficiaries of research in machine learning and AI, and also inspire foundational developments in it.

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.

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.

Recently proposed computational approaches explore casual links between chromatin and transcriptional changes that are provided by single-cell multimodal sequencing to bridge the knowledge gap in transcriptional regulatory control.

Genome architecture mapping (GAM) enables understanding of 3D genome structure in the nucleus. We directly compared multiplex-GAM and Hi-C data and found that local chromatin interactions were generally detected by both methods, but active genomic regions rich in enhancers that established higher-order contacts were preferentially detected by GAM.

Inferring gene networks from discrete RNA counts across cells remains a complex problem. Following Bayesian non-parametrics, a computational framework is proposed to perform non-biased inference of transcription kinetics from single-cell RNA counting experiments.