24 April 2025 in Communications Earth & Environment
Urbanization is revealed to cause more frequent and hotter day-night compound heatwaves, yet its impact on their onset remains unclear. This study compares summertime compound heatwaves in 722 urban-rural station pairs globally on an event-by-event basis during 1971-2017 and attributes their difference to numerous urbanization and climatological factors using interpretable machine learning methods. Here we find a discernible earlier onset (0.23 ± 0.04 days) of compound heatwaves in cities compared to rural areas. The magnitude of this influence is primarily determined by urban building volume and height rather than the artificial impervious surface fraction, and is exacerbated by climates characterized by warm night and ample daytime solar radiation. Taller buildings significantly advance onset in cloudy, warm, and humid climates, while larger building volumes lead to earlier onset of compound heatwaves across nearly all climates. Our findings suggest that cities, particularly those with large building volumes, should issue heatwave warnings earlier.
27 January 2022 in Geophysical Research Letters
Height‐time domain empirical orthogonal function (EOF) analysis is conducted to extract the dominant variations in the evolution of sudden stratospheric warming events (SSWs). The EOF1 corresponds to weak (I+) and strong (I−) polar vortex before the onset of SSWs and EOF2 indicates strong (W+) and weak (W−) warming during SSWs. The development of I− SSWs is associated with stronger eddy heat flux contributed by planetary wavenumbers 1 and 2, whereas the I + SSWs is related to wavenumber 1. W + SSWs are linked to both wavenumbers 1 and 2, while the happening of W− SSWs is associated with wavenumber 1. SSWs of I+ and W+ can exert influence on the surface such as negative phase of the North Atlantic Oscillation, which indicates that SSWs with different initial polar vortex states and warming intensities lead to different surface influence. This could help to better predict the tropospheric response following SSWs.
3 March 2026 in Nature Communications
Although soil moisture-atmosphere feedback (SAF) has been widely recognized as a key driver of temperature extremes, its distinct impacts on dry and humid heatwaves remain insufficiently understood, largely due to previous studies focusing predominantly on dry heatwaves alone. In this study, we combine SAF perturbation experiments from CMIP6 with partial differential analysis to construct a global map of SAF’s influence on both humid and dry heatwaves during 1985-2014 and to elucidate the underlying mechanisms. Unlike the homogeneously amplifying effect of SAF on dry heatwaves, SAF reduces the duration and severity of humid heatwaves by 20-40% in low-to-mid latitude regions with strong soil moisture-atmosphere coupling. In contrast, it increases their duration and severity by ≥50% in high latitude regions. These spatially divergent responses are primarily attributed to the competition between SAF-induced thermal warming and moisture depletion, which respectively raise or lower wet-bulb temperatures. Our findings reveal a dual regulatory mechanism by which SAF influences heat extremes through temperature-humidity coupling, providing scientific insights for region-specific adaptation strategies to both dry and humid heatwaves.
6 January 2025 in GigaScience
Cryogenic electron microscopy (cryoEM) has revolutionized structural biology by enabling atomic-resolution visualization of biomacromolecules. With artificial intelligence (AI) increasing role in newly developed cryoEM tools, task-specific datasets have become essential. Yet assembling such datasets often demands considerable effort and domain expertise, constraining AI-driven cryoEM tool development efforts. Here, we present CryoDataBot, an automated pipeline that addresses this gap. CryoDataBot streamlines data retrieval, preprocessing, and labeling, with fine-grained quality control and flexible customization, enabling efficient generation of robust datasets. CryoDataBot’s effectiveness is demonstrated through improved training efficiency in U-Net models and rapid, effective retraining of CryoREAD, a widely used RNA modeling tool. By simplifying the workflow and offering customizable quality control, CryoDataBot enables researchers to easily tailor dataset construction to the specific objectives of their models, while ensuring high data quality and reducing manual workload. This flexibility supports tools development for a wide range of applications in AI-driven structural biology.
10 September 2025 in bioRxiv
Cryogenic electron microscopy (cryoEM) has revolutionized structural biology by enabling atomic-resolution visualization of biomacromolecules. To automate atomic model building from cryoEM maps, artificial intelligence (AI) methods have emerged as powerful tools. Although high-quality, task-specific datasets play a critical role in AI-based modeling, assembling such resources often requires considerable effort and domain expertise. We present CryoDataBot, an automated pipeline that addresses this gap. It streamlines data retrieval, preprocessing, and labeling, with fine-grained quality control and flexible customization, enabling efficient generation of robust datasets. CryoDataBot’s effectiveness is demonstrated through improved training efficiency in U-Net models and rapid, effective retraining of CryoREAD, a widely used RNA modeling tool. By simplifying the workflow and offering customizable quality control, CryoDataBot enables researchers to easily tailor dataset construction to the specific objectives of their models, while ensuring high data quality and reducing manual workload. This flexibility supports a wide range of applications in AI-driven structural biology.
4 November 2026 in bioRxiv
SUMMARY: Identifying the molecular components of cryoEM/cryoET maps of endogenous or poorly characterized complexes remains a major bottleneck. CryoID2 addresses this by integrating deep learning–based feature prediction with sequence prioritization and structural evaluation. Three 3D U-Nets extract structure classes, key-atom positions, and reduced residue/nucleotide types from the input map. These predictions support backbone tracing and the generation of short sequence queries that rank candidates from a user-supplied pool—using BLASTp for proteins and a maximin strategy for RNA. Corresponding PDB or AlphaFold models are then retrieved, rigidly fitted using predicted features and experimental density, and ranked by map agreement. Module-level tests show high-recall feature prediction and resolution-dependent atom clustering, while a 52-case benchmark recovers near-reference fitting quality. In practice, cryoID2 correctly identified PfA-M17 from a 166-protein mass-spectrometry pool and the unannotated Ca-raiA RNA from a genome-derived non-coding sequence set, demonstrating a unified sequence-plus-structure route to molecular identification in cryoEM maps.