Comment

In this Comment, the authors overview the latest deep learning models for predicting regulatory function from genomic sequence and highlight key topics going forward, including the trade-off between specialized and general models, multitasking across cell types, and training on genetic variation and diverse species.

In this Comment, the authors outline some key next steps to advance our understanding of cis-regulatory elements at single-cell resolution, which includes harmonizing global efforts to construct a comprehensive single-cell atlas of gene regulation.

After more than two decades of large-scale efforts to annotate the regulatory genome, Sushant Kumar and Mark Gerstein forecast how new technologies and experimental approaches will pave the way in mapping regulatory elements across cell types, developmental stages and genetically diverse individuals.

In this Comment, Wendy Bickmore discusses mechanistic models of how 3D genome organization facilitates communication between distant enhancers and their target promoters to regulate gene expression.

Studies of human regulatory genomics are being performed at biobank scales, with data from tens of thousands of individuals. Stephen Montgomery describes how these datasets will advance our understanding of how variation in gene regulation shapes human traits and disease.

Transdisciplinary collaboration fuels innovation and discovery. Meller et al. call for broader collaboration at the intersection of genomics, the humanities and social sciences, and wider societal stakeholders, to test new ways of working across disciplines and co-develop future research agendas.

The accuracy of polygenic scores (PGS) remains limited and poorly transferable across ancestries. In this Comment, Zeng and Visscher discuss how integrating functional annotations with whole-genome sequencing data can improve PGS by prioritizing likely causal variants shared across populations and by assigning greater weight to variants in biologically relevant regions.

Prompt-based methods, which involve the careful design of inputs to guide large language model (LLM) outputs, are beginning to reshape bioinformatic analytical workflows. The authors compare prompt-driven approaches to conventional bioinformatics pipelines, outline their potential for multi-omics analysis and explore how these models may shape the future of computational biology.

Extreme environmental conditions create stressors that can interact with genetic risk factors to influence health outcomes. In this Comment, the authors discuss their vision for a national programme in Kuwait that combines the genome and exposome to uncover gene–environment interactions and inform tailored disease-prevention strategies.

Despite their immense potential, gene and cell therapies that target rare diseases are at risk of market withdrawal, owing to several challenges. The authors describe these hurdles and call for innovative measures to improve the economic sustainability of gene and cell therapies after regulatory approval.

Biologically informed neural networks promise to lead to more explainable, data-driven discoveries in genomics, drug development and precision medicine. Selby et al. highlight emerging opportunities, as well as challenges that will need to be overcome to enable their wider adoption.

Reflecting on the core values of early data sharing agreements, the Bermuda Principles and the Fort Lauderdale Agreement, Kathryn E. Holt and Michael Inouye emphasize the need to reaffirm our commitment to genomic data sharing to shape the future of science.

In this Comment, Agustín Robles-Remacho and Mats Nilsson highlight the opportunities and challenges of using spatial transcriptomics to detect and localize microRNAs in biological tissues, and advocate for the increased development of existing spatial transcriptomics methods.

The promise of paediatric genomics depends on proactively addressing complex ethical and equity issues with sustained community engagement. Hernandez et al. advocate for the integration of ELSI scholars into paediatric genomic study teams to catalyse timely discovery in genomics.

Gene therapy for congenital deafness has made a breakthrough in recent clinical trials. However, more work is needed to develop successful treatments for hereditary hearing loss, as the authors argue in this Comment.

In this Comment, the authors showcase how the emerging field of museomics — the application of multi-omic tools to natural history collections — is undergoing a rapid and exciting transformation, with new opportunities, challenges and confrontations with past legacies.

Rahimzadeh et al. discuss the ethical, legal and social implications of storing and analysing human genomic data in the cloud and provide recommendations and new research directions for future, trustworthy cloud-based genomic data access and management.

The adoption of microfluidics was fundamental to the development of cost-effective, high-throughput DNA sequencing. As the field progresses towards multi-omics, Lambert et al. reflect on the key concepts underlying microfluidics and how resulting engineering advances at the microscale drove the evolution of genomic sequencing.

In this Comment, Veltman and Tüttelmann call on geneticists to further study male infertility and help to develop diagnostic strategies using state-of-the-art genomic approaches.

Thirty years after the discovery and cloning of the cancer susceptibility gene BRCA1, William Foulkes reflects on this defining moment for breast and ovarian cancer genetics and how far the field has come.