Recent developments in imaging facilitate large-scale three-dimensional (3D) neuronal recording. While the resulting large datasets shed light on population-level neural coding, extracting neuronal calcium dynamics from 3D volumes remains more challenging than from two-dimensional images due to noise and scattering. Here we present DeepWonder3D, a general end-to-end pipeline for rapid and robust 3D neuronal extraction with high fidelity. Instead of processing voxel by voxel, DeepWonder3D works on the multiview projections of 3D imaging data obtained either digitally or optically through specific point spread functions and is therefore applicable to diverse techniques, including point-scanning microscopy, light-field microscopy and two-photon synthetic aperture microscopy. Integrating denoising, resolution registration, background removal, neuronal extraction and multiview fusion into a unified pipeline tailored for large-scale high-resolution datasets contaminated by noise and scattering, DeepWonder3D outperforms state-of-the-art methods in 3D localization accuracy with a tenfold reduction in computational costs, validated by numerical simulations and a hybrid two-photon/light-field imaging system. With the RUSH3D mesoscope, DeepWonder3D achieves high-fidelity 3D calcium extraction of tens of thousands of neurons across the mouse cortex within hours. DeepWonder3D is a pipeline for neuronal extraction from volumetric calcium imaging data acquired with a variety of one-photon or two-photon microscopy modalities.
Graphical abstracts are visual summaries that help communicate scientific discoveries.
Either I’ve been missing something or nothing has been going on. —Karen Elizabeth Gordon
A transparent evaluation and proof of reproducibility, generalization and replicability of algorithms are the bedrock of method development in computational biology. Many benchmarking efforts have been developed for problems ranging from structural biology to translational biomedicine. Rigor is relatively controllable for tasks such as the prediction of patient outcomes or the outcomes of biological assays, but the problem is exacerbated when the aim is to benchmark foundation models. The parameters constituting them are supposed to capture the patterns underlying the data; therefore, the models are parameterized embodiments of the phenomena that gave rise to the data. How can we test the limitations of these models? Here, we discuss the epistemological value of foundation models; whether they can be refuted, verified or evaluated primarily on the basis of utility; what principles should guide their benchmarking; and what role the scientific community should play in that benchmarking process.
Spatial long-read technologies are increasingly common but usually lack single-cell resolution. This leaves unanswered whether spatially variable isoforms reflect variability within one cell type or differences in region-specific cell-type composition. Here, we developed Spl-ISO-Seq2 (500-nm resolution) and accompanying software, Spl-IsoQuant-2 and Spl-IsoFind, enabling long-read sequencing of >450 million barcodes versus 80,000 previously. Applying this to the adult mouse brain, we compared differential isoform abundance between known regions and spatial isoform patterns independent of predefined regions. Both identified overlapping hits, for example, Rps24 in oligodendrocytes. For known Snap25 spatial isoform variation, we show that it occurs in excitatory neurons. The region-agnostic approach also uncovered patterns missed by region-based comparisons, for example, for Ighm. Notably, many spatial isoform signals are not driven by cell-type composition alone. Finally, our software is applicable to many spatial and single-cell protocols, demonstrating reproducibility between platforms (for example, Visium HD/Stereo-seq). Overall, our experimental/analytical methods enable a submicron-resolution-isoform view and open avenues for spatial isoform disease research.
Antigen recognition by T cells through their highly diverse T cell receptors (TCRs) is central to adaptive immunity, and recent high‑throughput methods and databases have greatly expanded the availability of TCR–antigen specificity data. However, much of this information is noisy or weakly validated. We explore the field’s main bottlenecks, which are data quality rather than quantity, motivating rigorous experimental validation, statistical filtering and multimodal AI approaches to derive high-confidence TCR specificity records.
Organoids are transforming biomedical science, but their promise hinges on shifting from episodic ethics review to sustained, integrated stewardship. We argue that such stewardship must evolve alongside the science to responsibly honor the human origins of these innovative models.
Cellular function depends on the spatial organization of cells and biomolecules within the tissue microenvironment. Advances in spatial omics have enabled profiling of molecular features such as transcriptome, proteome and epigenome, and there has been rapid progress of imaging-based approaches to study spatial three-dimensional (3D) genome organization. Here we present Spatial-ATAC-Hi-C, a microfluidic‑based platform for genome-wide, spatially resolved joint-profiling of 3D genome organization and chromatin accessibility on tissue slides. Applied to mouse and human brains, Spatial-ATAC-Hi-C revealed distinct chromatin architecture and gene regulatory programs in neuronal and non-neuronal populations in their native tissue context. In glioblastoma and astrocytoma samples, we detected spatially resolved 3D genome alterations, copy number variations and structural variations across tumor regions, revealing clinically relevant oncogenic events and clonal heterogeneity. By co-profiling of genome architecture and chromatin accessibility while preserving tissue architecture, Spatial-ATAC-Hi-C provides a powerful tool for studying spatial gene regulation in human biology and disease.
Re-analysis of published long non-coding RNA (lncRNA) screens has revealed unexpectedly high false-positive rates in some reports, indicating that high-throughput genetic perturbations are more prone to methodological artifacts when applied to lncRNAs than to protein-coding genes. We highlight common technical and analytical sources of false positives in lncRNA screens and outline strategies to detect, monitor and mitigate these issues.
Fluorescence super-resolution microscopy has advanced optical imaging into the nanoscale regime, transforming biological and interdisciplinary research. However, wide-field super-resolution techniques often compromise temporal resolution, thereby limiting the capture of rapid and transient biological events in living systems. Here we introduce spatial polarization-induced fluorescence fluctuation imaging (SPIFFI), a multichannel polarimetric method for single-shot super-resolution imaging and six-dimensional information extraction. By leveraging the inherently smaller point spread function under polarized detection and capturing polarization-dependent spatial fluctuations across multiplexed channels, SPIFFI achieves instant resolution enhancement from a single exposure. This capability substantially enhances the feasibility of volumetric live-cell super-resolution imaging. Moreover, SPIFFI images can integrate seamlessly with existing fluctuation-based methods for further postprocessing and resolution improvement. We demonstrate the versatility of SPIFFI through experiments on both fixed and live cells, capturing rapid subcellular dynamics and enabling high-throughput, multidimensional imaging beyond the diffraction limit. SPIFFI thus offers a practical and robust platform for real-time super-resolution imaging in biological research.
In this Comment, we direct attention to initial efforts to establish a high-quality databank of atomic force microscopy (AFM) data: bioAFM-DB. We outline the state of this endeavor, its challenges, and potential courses of action.
Specialized research software has historically been costly and time-consuming to create, but large language models (LLMs) have become capable enough at code generation to fundamentally change this. We describe how LLM-assisted programming disrupts the landscape by allowing researchers to build tools without support from software engineers, illustrate this with an example built rapidly by a single LLM-assisted developer, and discuss opportunities and risks.
