
Here we introduce the AMP Challenge, a NeurIPS 2026 competition that provides the first standardized, blinded, head-to-head benchmark for generative AI models designing anti-microbial peptides, addressing the field’s urgent need for a shared way to identify which computational strategies actually translate into real antimicrobial activity. Antimicrobial resistance is projected to become one of the leading causes of death worldwide, having already contributed to an estimated 4.95 million deaths in 2019 [1]
Asia contains over 60% of the global population and vast genetic diversity, yet its genomic resources remain underrepresented. Over the past decade or so, regional research has evolved from fragmented surveys into large-scale national precision medicine initiatives. Here, we examine how these sequencing efforts collectively form an emerging Asian genomic infrastructure. We discuss future directions to close critical gaps in Central Asia, minority populations, cross-project interoperability, and equitable access, highlighting the transition toward interoperable, clinical-grade genomic ecosystems supporting precision medicine across the continent.
Abstract Background G-quadruplexes (G4s) are non-canonical DNA structures that have gained increasing attention for their potential roles in gene regulation and role in neurodegenerative diseases and cancer. Despite their biological significance, G4 structures have not been studied systematically across tissues and cell types. Results In this study, we employ G4 single-cell CUT&Tag (G4 scCUT&Tag) to characterize G4 landscapes in postnatal mouse brain cells, leveraging single-cell analytical approaches commonly used in scRNA-Seq and scATAC-Seq datasets. Using conventional single-cell omics workflows to process and explore our data, we distinguish distinct cell lineages based on G4 heterogeneity and establish that a subset of lineage-specific genes show unique promoter G4s. Multi-omics integration with scRNA-Seq gene expression profiles, using both a covariance-based technique (canonical correlation analysis) and a transfer learning-based approach, enables more accurate cell-type annotation. These integrations not only reveal correlations between G4 signal and gene expression, but demonstrate that G4 scCUT&Tag enables detailed examination of G4 heterogeneity in complex tissues and supports integrative analysis of G4 profiles with other omics layers, offering new insights into the epigenomic landscapes of the developing brain. Conclusions Our findings establish G4 scCUT&Tag as a framework for investigating the cellular heterogeneity of G4 landscapes in complex tissues and provide a basis for exploring their relationships with gene regulation and other layers of chromatin organization.
Eukaryotic Fanzor (Fz) proteins are compact RNA-guided endonucleases homologous to prokaryotic TnpB/Cas12, offering theoretical advantages for single AAV delivery. However, their low editing efficiency and poorly defined specificity in mammals have hindered therapeutic development. Here, we perform structure-guided engineering of Spizellomyces punctatus Fanzor1 nuclease (SpuFz1) to enhance its utility for in vivo genome editing and identify several Fanzor variants whose efficiencies are comparable or only slightly lower than widely-used miniature enAsCas12f1 and large-sized SpCas9. We demonstrate that catalytically dead dSpuFz1-v4.2 can be repurposed for CRISPRa-mediated transcriptional activation in a proof-of-concept manner. Genome-wide PEM-seq in mice reveal low translocation frequencies (below 0.05
The pig (Sus scrofa) is widely studied as a large-animal model for human gastrointestinal biology, including gut physiology, pathology and disease. However, the spatial and developmental organisation of the pig intestine is not yet fully understood. Here, we generate a comprehensive spatiotemporal atlas of pig intestinal development by applying spatial transcriptomics to three intestinal segments across five developmental stages and an additional four intestinal segments at the adult stage. We identify spatial separation among enterocytes, plasma cells and stem cells, observing more pronounced separation in the small intestine compared to the large intestine. Pseudotime analysis of enterocyte-enriched regions reveals a gradient along the crypt-villus/crypt axis as early as post-conception day 70 in the jejunum, becoming fully evident across all intestinal segments by birth in the large intestine, with all three intestinal segments reaching a mature state by postnatal day 33. Our analysis indicates that the maturation of enterocytes during migration along the crypt-villus axis involves transcriptional programs similar to those governing enterocyte maturation across developmental stages, suggesting conserved molecular mechanisms governing both spatial and temporal differentiation. In the large intestine, nutrient transport-related gene expression declines progressively during development, transitioning from jejunum-like levels at early stages to markedly reduced levels in adulthood, consistent with functional specialization of distinct intestinal segments. Our study provides a high-resolution spatiotemporal framework of pig intestinal development and establishes a valuable resource for comparative biology and translational research.