Cell segmentation is a critical yet under-standardized step in spatial transcriptomics, directly affecting cell identity assignment, spatial organization and downstream biological interpretation. Despite rapid advances in spatial transcriptomics technologies, systematic evaluations of segmentation methods across platforms and biological contexts remain limited. Here we benchmark 11 image-based, transcriptomics-based and multimodal segmentation methods across 15 datasets generated by 5 spatial transcriptomics technologies, spanning 12 human and mouse tissue types. We introduce biologically informed evaluation metrics that quantify segmentation quality, transcript recovery, cell composition, morphology, expression purity and spatial organization. Benchmarking reveals substantial variation in performance across datasets, technologies and evaluation criteria, highlighting trade-offs among segmentation completeness, transcriptomic fidelity and biological interpretability. Cross-technology analyses further demonstrate that segmentation choice critically shapes downstream analyses, including cell-type specificity, transcript contamination and ligand–receptor inference. Together, our unified Nextflow pipeline provides a scalable foundation for standardized evaluation of segmentation and method selection in spatial transcriptomics. This Analysis benchmarks segmentation methods across different datasets and presents Nextflow, a pipeline for standardized method selection for spatial transcriptomics.
Biological systems exhibit complex hierarchical regulation, with interactions across multiple layers governing transitions between states. Identifying the causal architecture underlying these state transitions remains challenging because confounding can produce spurious associations in observational data. Here we present CauTrigger, a hierarchical deep latent variable model that integrates do-calculus to identify potential causal triggers and predict perturbation outcomes across multiple regulatory layers. By disentangling causal and spurious latent factors, CauTrigger recursively reconstructs the underlying regulatory hierarchy and enables interpretable reasoning from observational data. Across synthetic and real-world datasets with multilayer regulatory structures, CauTrigger outperforms existing methods in identifying causal triggers and forecasting system-level responses. Applied to single-cell multi-omics and gut microbiome datasets, CauTrigger prioritizes candidate causal triggers and uses in silico perturbations to generate testable hypotheses about biological state transitions and inform potential strategies for state navigation. Together, these results establish CauTrigger as a general-purpose framework for causal discovery in complex biological systems. CauTrigger is a hierarchical deep latent variable model that can identify potential causal triggers and predict outcomes of perturbation events.
Since the discovery and application of green fluorescent protein for use as an optical tag, fluorescent proteins (FPs) have changed the way we observe biological signals and processes. FP and fluorescent biosensor design has continued to advance, facilitated by new technologies and developments in chemistry, physics, engineering and screening technologies. We summarize some of the recent developments in FP engineering, focusing on insights that have led to improved photostability or longer wavelengths, among other improvements, highlighting mScarlet as a case study. We also discuss the engineering of new fluorescent biosensors, focusing on the ways in which screening, quantitative imaging, spatial resolution and multiplexing have been improved in recent years. Finally, we conclude with a perspective on what the future holds for the next generation of FPs and biosensors. This Review summarizes recent developments in fluorescent protein technology, including the engineering of fluorescent biosensors.
Here we present AmpliconSuite, a toolkit for focal amplification analysis from cancer whole-genome sequencing that identifies extrachromosomal DNA (ecDNA), breakage–fusion–bridges and other amplification modes. Its ecContext module categorizes ecDNA structures to understand formation mechanisms. Validation across 175 metaphase fluorescence in situ hybridization-probed genes in 72 models demonstrated accurate predictions. Analysis of 6,506 patient pan-cancer samples revealed ecDNA-mediated oncogene co-amplification preferences. To democratize data sharing, we created AmpliconRepository, a community-contributable platform, enabling investigation of results across >15,000 publicly hosted samples. AmpliconSuite is a toolkit that enables the analysis of focal amplifications from cancer whole-genome sequencing data.
Addressing sex-related variation might seem like an experimental or data analysis time-sink. It’s not.
