Land-surface models commonly assume that soil CO2 production is rapidly transferred out of the soil profile, effectively equating respiration with near-surface signals. However, persistent subsoil CO2 enrichment despite limited in situ production reflects bidirectional exchange of topsoil-respired CO2, suggesting that the surface-centric assumption may not be universally valid. Here we combine year-round, multi-depth field observations with process-based modeling (DeepCO2) to quantify the effects of land use (croplands, forests, and forest-cropland edges) and climate on soil CO2 production-transport dynamics. We reveal pronounced vertical decoupling of soil CO2, with deep CO2 (> 1 m) concentrations up to two orders of magnitude above surface concentrations and 9-17 times greater than can be explained by in situ production alone, revealing profile-scale production-transport imbalance. Forest soils promote ~2.3-fold greater deep CO2 accumulation than cropland and edge systems, driven by enhanced macroporosity and stronger advective coupling. Seasonal climate forcing regulates these regimes, with winter cold spells suppressing exchange, whereas spring-summer warming amplifies CO2 source strength originating from otherwise "protected" subsoil organic carbon, particularly in cropland and edge soils. These findings demonstrate that a substantial fraction of topsoil-respired CO2 is redistributed below the soil-atmosphere boundary and retained within the subsurface, forming a land-use-sensitive CO2-C sink. Surface-based CO2 assessment therefore reflects a transport-limited, boundary-conditioned signal rather than a direct proxy for profile-scale production, and alone cannot resolve the full partitioning of soil-respired carbon. Incorporating depth-dependent production-transport coupling is therefore essential for accurately representing terrestrial carbon allocation and land-climate feedbacks under ongoing land-use change and climate warming.

Fungal phytopathogens pose a persistent threat to global crop production, and widespread use of fungicides has driven resistance development and environmental concerns, necessitating sustainable alternatives. Actinomycetes produce diverse bioactive metabolites, yet natural product discovery has disproportionately focused on Streptomyces, leaving rare actinomycete taxa underexplored. Saccharomonospora xinjiangensis XJ-54 is a rare actinomycete exhibiting antifungal activity against Fusarium phytopathogens, including Fusarium oxysporum f. sp. cucumerinum (FORC), and harbors numerous biosynthetic gene clusters (BGCs) whose products remain uncharacterized. However, as in many rare actinomycetes, BGCs may be transcriptionally silent under standard laboratory conditions, and their expression dynamics remain poorly understood. To elucidate the molecular basis of antifungal activity in S. xinjiangensis XJ-54, we integrated genomic, transcriptomic, and metabolomic analyses. Cell-free supernatants inhibited FORC after 5 days of fermentation, with activity increasing by day 7. Time-resolved RNA sequencing demonstrated that all genomically identified BGCs were transcriptionally active but exhibited distinct growth-phase-dependent expression patterns, with approximately half upregulated during exponential growth and the remainder following transition to stationary phase. We observed temporal variations in transcriptional coupling between cluster-specific regulators and biosynthetic genes. Liquid chromatography-mass spectrometry-based metabolomics showed growth-phase-dependent metabolite shifts, including stationary phase accumulation of secoiridoid-like monoterpenoids, N-acyl amines, and alkaloids (imidazoles, pyridines, and indoles), correlating with the observed antifungal phenotype. Bioactivity-guided fractionation subsequently yielded an active fraction containing a predicted halogenated alkaloid that induced hyphal damage in FORC. These findings indicate that antifungal activity in S. xinjiangensis XJ-54 arises from temporally coordinated biosynthetic programs, providing a framework for optimizing growth conditions and prioritizing BGCs for functional characterization.IMPORTANCEThe discovery of new antifungal compounds is critical for sustainable agriculture, yet most natural product research has focused on a narrow range of well-studied microorganisms. Rare actinomycetes represent an untapped reservoir of chemical diversity, but their biosynthetic potential often remains hidden. By integrating time-resolved transcriptomics with metabolomics, we show that the rare actinomycete Saccharomonospora xinjiangensis XJ-54 produces antifungal metabolites through temporally coordinated biosynthetic programs. Contrary to the prevailing assumption that the majority of biosynthetic gene clusters (BGCs) are silent, all BGCs in this strain showed detectable transcriptional activity under standard laboratory conditions, with expression patterns that were strongly growth-phase dependent. These findings provide a framework for linking gene cluster expression to metabolite production and bioactivity, enabling more rational prioritization of BGCs and optimization of cultivation strategies. More broadly, this work provides a roadmap for unlocking the biosynthetic potential of rare actinomycetes and accelerating the discovery of antifungal natural products relevant to sustainable crop protection.

Insulin resistance is conventionally framed as a pathological defect in insulin action that contributes to type 2 diabetes and related metabolic disease and should therefore be identified and reversed. The comprehensive review by Gastaldelli and colleagues highlights a more complicated reality: insulin resistance is heterogeneous between individuals, tissues, pathways, and physiological states. I argue that this heterogeneity exposes a fundamental issue-insulin resistance is a phenotype, not a mechanism. It can arise through many molecular routes, occur in both disease and normal physiology, originate in different tissues, and long precede overt disease. I propose an "insulin resistance mosaic" in which its clinical significance depends on why and where insulin resistance develops, physiological compensation, environmental exposure, and end-organ vulnerability. Similar degrees of insulin resistance could therefore lead to very different clinical outcomes, remain compensated for decades, or even, in some contexts, be protective. This framework challenges the assumption that insulin resistance represents a single disease process that should invariably be overcome. Instead, we should determine why it has developed, where it resides, what it predicts, and whether modifying it improves clinical outcomes.

Insulin resistance is conventionally framed as a pathological defect in insulin action that contributes to type 2 diabetes and related metabolic disease and should therefore be identified and reversed. The comprehensive review by Gastaldelli and colleagues highlights a more complicated reality: insulin resistance is heterogeneous between individuals, tissues, pathways, and physiological states. I argue that this heterogeneity exposes a fundamental issue-insulin resistance is a phenotype, not a mechanism. It can arise through many molecular routes, occur in both disease and normal physiology, originate in different tissues, and long precede overt disease. I propose an "insulin resistance mosaic" in which its clinical significance depends on why and where insulin resistance develops, physiological compensation, environmental exposure, and end-organ vulnerability. Similar degrees of insulin resistance could therefore lead to very different clinical outcomes, remain compensated for decades, or even, in some contexts, be protective. This framework challenges the assumption that insulin resistance represents a single disease process that should invariably be overcome. Instead, we should determine why it has developed, where it resides, what it predicts, and whether modifying it improves clinical outcomes.

Logistic regression is a statistical method used for exploratory and predictive purposes in clinical research. While it is widely used, logistic regression results reported across studies have been inconsistent, particularly in terms of model estimation, validation, and interpretation, even though most studies have clarified their analytical purposes. Reports frequently do not adequately describe how the logistic regression models were evaluated in a purpose-specific manner, particularly regarding underlying assumptions, diagnostic procedures, and performance assessments. Therefore, the interpretability, reproducibility, and clinical applicability of the results presented are frequently limited. We reviewed papers published in the Korean Journal of Anesthesiology (KJA) between 2021 and 2024 that used logistic regression and examined reporting practices, focusing on whether the model evaluation was consistent with the stated analytical purpose. Based on our findings, we then outline the core assumptions required for the appropriate use of logistic regression and introduce key diagnostic tools with practical guidance on their application according to analytical purposes. We also describe statistically robust approaches for reporting the results of logistic regression analyses used for exploratory and predictive purposes, as well as study designs that intentionally incorporate both these aims. With this article, we hope to help researchers establish clear analytical objectives and to adopt systematic reporting practices when using logistic regression, thereby strengthening the basis for research findings to serve as scientifically credible evidence for clinical decision-making.

There is growing interest in a hybrid control design for treatment evaluation, where a randomized controlled trial is augmented with external control data from a previous trial or a real-world data source. The hybrid control design has the potential to improve efficiency but also carries the risk of introducing bias. The potential bias in a hybrid control study can be mitigated by adjusting for baseline covariates that are related to the control outcome. Existing methods that serve this purpose commonly assume that the internal and external control outcomes are exchangeable upon conditioning on a set of measured covariates. Possible violations of the exchangeability assumption can be addressed using a g-computation method with variable selection under a correctly specified outcome regression model. In this article, we point out that a particular version of this g-computation method is protected against misspecification of the outcome regression model. This observation leads to a model-robust g-computation method that is remarkably simple and easy to implement, consistent and asymptotically normal under minimal assumptions, and able to improve efficiency by exploiting similarities between the internal and external control groups. The method is evaluated in a simulation study and illustrated using real data from HIV treatment trials.

Missing data can pose fundamental challenges to statistical inference, with nonignorable missing or missing not at random (MNAR) presenting the most severe methodological difficulties. Despite substantial advances in MNAR inference methods, several limitations remain. These generally include interpretability issues, non-identifiability, unverifiable parametric assumptions, and reliance on external information for which clear practical guidance is often lacking-particularly in sensitivity and Bayesian methods. In this paper, motivated by a real-world COVID-19 dataset from the Centers for Disease Control and Prevention (CDC), we study an MNAR scenario in the presence of a shadow variable that may itself also be subject to MNAR. A shadow variable is associated with the primary variable that is MNAR, but is conditionally independent of that variable's missingness given the primary variable and other covariates. Under the pattern mixture model framework, we propose a nonparametric inference method that can leverage external data to explicitly address the identification and estimation problems in the primary data. The method requires inference only of an observed data density and an odds of missing function, with the latter forming the core of our approach. We further provide two multiple-imputation-based estimation strategies, enhancing transparency and interpretability. The proposed framework can accommodate both covariate MNAR and outcome MNAR settings, as well as different variable types, and extends the existing shadow variable paradigm for MNAR data by relaxing the common assumption that the shadow variable must be fully observed. We evaluate the proposed method through simulation studies and apply it to CDC COVID-19 surveillance data.

Anaerobic digestion (AD) plays a pivotal role in renewable energy generation and waste valorisation, yet its optimisation is constrained by the limitations of conventional modelling and control strategies. The Anaerobic Digestion Model No.1 (ADM1) remains the most widely adopted mechanistic model. However, its high structural complexity, defined by 26 state variables and over 100 parameters, presents significant calibration challenges. These rigid biological assumptions severely constrain the applicability of the model for real-time prediction, dynamic optimisation, and process control under transient loading. This review synthesises developments across three converging research domains: (i) ADM1 extensions, modifications, and reductions, including substrate‑specific kinetics, mass‑balanced formulations, and surrogate‑based simplifications; (ii) emerging MC modelling approaches, spanning genome‑scale metabolic reconstructions, and hybrid mechanistic-data‑driven frameworks; and (iii) advanced control strategies, including model‑predictive control, soft sensing, and AI‑driven or reinforcement learning for dynamic, uncertainty‑aware operation. The critical analysis highlights that future AD optimisation depends on integrating microbial ecological insight with computationally efficient models and data‑driven intelligence to produce robust, site‑adaptive decision‑support tools. We identify key knowledge gaps and propose how to move towards a unified modelling-microbial-control framework to accelerate deployment of intelligent AD systems to improve biogas yield, stability, and resilience at industrial scale.