Search

The rapid growth of large-scale spatial gene expression data demands efficient and reliable computational tools to extract major trends of gene expression in their native spatial context. Here, we used stability-driven unsupervised learning (i.e., staNMF) to identify principal patterns (PPs) of 3D gene expression profiles and understand spatial gene distribution and anatomical localization at the whole mouse brain level. Our subsequent spatial correlation analysis systematically compared the PPs to known anatomical regions and ontology from the Allen Mouse Brain Atlas using spatial neighborhoods. We demonstrate that our stable and spatially coherent PPs, whose linear combinations accurately approximate the spatial gene data, are highly correlated with combinations of expert-annotated brain regions. These PPs yield a brain ontology based purely on spatial gene expression. Our PP identification approach outperforms principal component analysis and typical clustering algorithms on the same task. Moreover, we show that the stable PPs reveal marked regional imbalance of brainwide genetic architecture, leading to region-specific marker genes and gene coexpression networks. Our findings highlight the advantages of stability-driven machine learning for plausible biological discovery from dense spatial gene expression data, streamlining tasks that are infeasible by conventional manual approaches., Competing Interests: Competing interests statement:The authors declare no competing interest.

Improving the scalability of tissue imaging throughput with bright, coherent X-rays requires identifying and mitigating artifacts resulting from the interactions between X-rays and matter. At synchrotron sources, long-term imaging of soft tissues in solution can result in gas bubble formation or cavitation, which dramatically compromises image quality and integrity of the samples. By combining in-line phase-contrast imaging with gas chromatography in real time, we were able to track the onset and evolution of high-energy X-ray-induced gas bubbles in ethanol-embedded soft tissue samples for tens of minutes (two to three times the typical scan times). We demonstrate quantitatively that vacuum degassing of the sample during preparation can significantly delay bubble formation, offering up to a twofold improvement in dose tolerance, depending on the tissue type. However, once nucleated, bubble growth is faster in degassed than undegassed samples, indicating their distinct metastable states at bubble onset. Gas chromatography analysis shows increased solvent vaporization concurrent with bubble formation, yet the quantities of dissolved gasses remain unchanged. By coupling features extracted from the radiographs with computational analysis of bubble characteristics, we uncover dose-controlled kinetics and nucleation site-specific growth. These hallmark signatures provide quantitative constraints on the driving mechanisms of bubble formation and growth. Overall, the observations highlight bubble formation as a critical yet often overlooked hurdle in upscaling X-ray imaging for biological tissues and soft materials and we offer an empirical foundation for their understanding and imaging protocol optimization. More importantly, our approaches establish a top-down scheme to decipher the complex, multiscale radiation-matter interactions in these applications., (open access.)