Proceedings of the National Academy of Sciences of the United States of America最新文献

The unsupervised and principled diagnosis of multiscale data is a fundamental obstacle in modern scientific problems from, for instance, weather and climate prediction, neurology, epidemiology, and turbulence. Multiscale data are characterized by a combination of processes acting along multiple dimensions simultaneously, spatiotemporal scales across orders of magnitude, nonstationarity, and/or invariances such as translation and rotation. Existing methods are not well-suited to multiscale data, usually requiring supervised strategies such as human intervention, extensive tuning, or selection of ideal time periods. We present the multiresolution coherent spatio-temporal scale separation (mrCOSTS), a hierarchical and automated algorithm for the diagnosis of coherent patterns or modes in multiscale data. mrCOSTS is a variant of dynamic mode decomposition which decomposes data into bands of spatial patterns with shared time dynamics, thereby providing a robust method for analyzing multiscale data. It requires no training but instead takes advantage of the hierarchical nature of multiscale systems. We demonstrate mrCOSTS using complex multiscale datasets that are canonically difficult to analyze: 1) climate patterns of sea surface temperature, 2) electrophysiological observations of neural signals of the motor cortex, and 3) horizontal wind in the mountain boundary layer. With mrCOSTS, we trivially retrieve complex dynamics that were previously difficult to resolve while additionally extracting hitherto unknown patterns of activity embedded in the dynamics, allowing for advancing the understanding of these fields of study. This method is an important advancement for addressing the multiscale data which characterize many of the grand challenges in science and engineering.

The introns of the gene encoding the long noncoding RNA (lncRNA) GAS5 host up to 10 C/D box small nucleolar RNAs (snoRNAs). However, whether there is a regulatory and functional relationship between these snoRNAs and GAS5 is unknown. Here, we show that the expression of SNORD80, but not the other snoRNAs, parallels GAS5 expression and is regulated alongside GAS5 in response to cellular stress. The 2'-O-methylation at the A496 site, located within a segment of GAS5 complementing the conserved RNA-binding region on SNORD80, promotes GAS5 stability and consequent upregulation. This methylation requires SNORD80, as it is diminished by knockdown of SNORD80 and increased by SNORD80 overexpression, similar to the effects of manipulating the expression of fibrillarin, the methyltransferase of the box C/D small nucleolar ribonucleoprotein particle (snoRNP). The upregulation of SNORD80 in response to cellular stress is due to an enhancement in its stability, which is associated with an increase in its interaction with fibrillarin. Collectively, these results identify a role for SNORD80 in guiding 2'-O-methylation to stabilize GAS5. This uncovers a feedforward regulatory loop at the GAS5 gene locus in response to cellular stress and sheds light on posttranscriptional mechanisms governing lncRNA expression.

In this study, we used an inhibitor of phosphodiesterase 6 (PDE6) to examine the impact of changes in the conformation of the PDE6 protein on the light-induced process responsible for altering the length of the outer segments of photoreceptor cells in both human and rodent eyes. We employed a imaging method called spatiotemporal optical coherence tomography, which ensures high contrast and phase stability within the strongly scattering photoreceptor- Retinal Pigment Epithelium complex. Using this approach, we recorded nanometer-scale changes in human cones and rods in response to photopic flicker stimulation and observed length changes in rodent rods under scotopic conditions following a single pulse of light, in the absence or presence of sildenafil, which inhibits the catalytic activity of PDE6. Our findings are consistent with the interpretation that during phototransduction conformational changes in PDE6 structure, which occur on an angstrom scale, are amplified to the nanometer scale due to the unique structure of the photoreceptor outer segments and sequential stimulation. This finding opens up possibilities for the informed use of photopic flicker optoretinography measurements as a diagnostic tool, as the observed nanometer-scale changes in rod and cone dimensions as a function of light stimulus can now be directly linked to molecular events involved in the phototransduction pathway.

Modern machine learning has achieved impressive prediction performance, but often sacrifices interpretability, a critical consideration in high-stakes domains such as medicine. In such settings, practitioners often use highly interpretable decision tree models, but these suffer from inductive bias against additive structure. To overcome this bias, we propose Fast Interpretable Greedy-Tree Sums (FIGS), which generalizes the Classification and Regression Trees (CART) algorithm to simultaneously grow a flexible number of trees in summation. By combining logical rules with addition, FIGS adapts to additive structure while remaining highly interpretable. Experiments on real-world datasets show FIGS achieves state-of-the-art prediction performance. To demonstrate the usefulness of FIGS in high-stakes domains, we adapt FIGS to learn clinical decision instruments (CDIs), which are tools for guiding decision-making. Specifically, we introduce a variant of FIGS known as Group Probability-Weighted Tree Sums (G-FIGS) that accounts for heterogeneity in medical data. G-FIGS derives CDIs that reflect domain knowledge and enjoy improved specificity (by up to 20% over CART) without sacrificing sensitivity or interpretability. Theoretically, we prove that FIGS learns components of additive models, a property we refer to as disentanglement. Further, we show (under oracle conditions) that tree-sum models leverage disentanglement to generalize more efficiently than single tree models when fitted to additive regression functions. Finally, to avoid overfitting with an unconstrained number of splits, we develop Bagging-FIGS, an ensemble version of FIGS that borrows the variance reduction techniques of random forests. Bagging-FIGS performs competitively with random forests and XGBoost on real-world datasets.

Unlike coffee and cream that homogenize when stirred, growing micro-organisms (e.g., bacteria, baker's yeast) can actively kill each other and avoid mixing. How do such antagonistic interactions impact the growth and survival of competing strains, while being spatially advected by turbulent flows? By using numerical simulations of a continuum model, we study the dynamics of two antagonistic strains that are dispersed by incompressible turbulent flows in two spatial dimensions. A key parameter is the ratio of the fluid transport time to that of biological reproduction, which determines the winning organism that ultimately takes over the whole population from an initial heterogeneous state, a process known as fixation. By quantifying the probability and mean time for fixation along with the spatial structure of concentration fluctuations, we demonstrate how turbulence raises the threshold for biological nucleation and antagonism suppresses flow-induced mixing by depleting the population at interfaces. Our work highlights the unusual biological consequences of the interplay of turbulent fluid flows with antagonistic population dynamics, with potential implications for marine microbial ecology and origins of biological chirality.

Acute lymphoblastic leukemia (ALL) poses challenges in adult patients, considering its heterogeneous nature and often suboptimal treatment outcomes. Here, we performed a study on 201 newly diagnosed adult ALL cases (age ≥ 15 y) to generate intracellular and dynamic serum metabolomic profiles. Our findings revealed a predominant increase in bile acid (BA) metabolites in serum, alongside metabolic rewiring that supported highly proliferative states and actively metabolic signaling, such as enriched nucleotide metabolism in leukemic blasts. By integrating intracellular metabolomics and transcriptomics, we constructed the Comprehensive Metabolic Information Dataset (CMID), which facilitated the development of a clustering system to supplement current risk stratification. Furthermore, we explored potential metabolic interventions targeting the serum BA profile and energy metabolism in blasts. The combined use of simvastatin with vincristine and dexamethasone regimen demonstrated a synergistic therapeutic effect in a murine ALL model, effectively lowering key BA levels in serum and suppressing the infiltration of leukemic blasts in the liver. In light of the enhanced intracellular redox metabolism, combining FK866 (a nicotinamide phosphoribosyltransferase inhibitor) and venetoclax significantly prolonged survival in a patient-derived xenograft ALL model. Our findings, along with the resulting resources (http://www.genetictargets.com/MALL), provide a framework for the metabolism-centered management of ALL.