https://3demmethods.i2pc.es/index.php?action=history&feed=atom&title=2026Sun_QwenCryoMarker 2026-09-04T06:55:40Z Revision history for this page on the wiki MediaWiki 1.44.2 https://3demmethods.i2pc.es/index.php?title=2026Sun_QwenCryoMarker&diff=5275&oldid=prev 2026-09-03T05:57:51Z

<p>Created page with &quot;== Citation == Sun, Y., Zhao, J., Xu, N., Wang, L., Ding, W. and Li, M. 2026. QwenCryoMarker: a universal post-processing framework for contamination-aware particle cleaning. Acta Crystallographica Sec. D. 82, 9 (2026). == Abstract == Cryo-electron microscopy (cryo-EM) micrographs are frequently contaminated by carbon edges, ice crystals, ethane bubbles and other high-contrast artifacts. These contaminants trigger abundant false positives in automated particle pickers...&quot;</p> <p><b>New page</b></p><div>== Citation ==<br /> <br /> Sun, Y., Zhao, J., Xu, N., Wang, L., Ding, W. and Li, M. 2026. QwenCryoMarker: a universal post-processing framework for contamination-aware particle cleaning. Acta Crystallographica Sec. D. 82, 9 (2026).<br /> <br /> == Abstract ==<br /> <br /> Cryo-electron microscopy (cryo-EM) micrographs are frequently contaminated<br /> by carbon edges, ice crystals, ethane bubbles and other high-contrast artifacts.<br /> These contaminants trigger abundant false positives in automated particle<br /> pickers, severely hampering downstream 3D reconstruction. Existing methods<br /> either avoid contamination implicitly (requiring dataset-specific tuning) or rely<br /> on rule-based filters that fail on complex contamination patterns. Here, we<br /> present QwenCryoMarker, a universal post-processing framework that converts<br /> the outputs of arbitrary particle pickers into clean, high-precision particle sets.<br /> Our pipeline consists of two core stages: (i) a visual large model (Qwen-Image-<br /> Edit-2511) fine-tuned via supervised learning to generate pixel-accurate binary<br /> contamination masks from raw micrographs and (ii) a lightweight contaminationaware<br /> filtering module that discards particles when the contamination proportion<br /> within their surrounding circular region exceeds a predefined threshold<br /> ratio. Contamination is defined as all micrograph regions unsuitable for reliable<br /> particle picking and subsequent 3D reconstruction, including carbon edges, ice<br /> crystals, ethane bubbles, miscellaneous debris and dense protein aggregates.<br /> The framework features plug-and-play deployment: it requires no per-dataset<br /> parameter tuning or extra retraining, and maintains compatibility with classical<br /> pickers (blob detection, template matching) as well as deep learning-based<br /> pickers (Topaz, crYOLO etc.). We validate QwenCryoMarker on five diverse<br /> CryoPPP benchmark datasets across four representative particle pickers.<br /> Quantitatively, our method consistently boosts precision with an average<br /> absolute gain of 0.009 and lifts the F1-score, while recall only drops slightly by an<br /> average of 0.008. Segmentation benchmarking shows our model reaches a mean<br /> intersection over union (IoU) of 0.629, surpassing that of the state-of-the-art<br /> MicrographCleaner (0.551) by 14.2%. We further compare against multiple<br /> segmentation baselines: U-Net (0.448), DeepLabV3+ (0.488), SAM (0.475),<br /> ASOCEM (0.195) and IceBreaker (0.433). A downstream reconstruction case<br /> study on EMPIAR-10017 verifies that particles filtered by QwenCryoMarker<br /> yield cleaner 2D class averages and higher resolution 3D density maps (3.88<br /> versus 3.97 A ˚ ). Qualitative visualization also confirms that QwenCryoMarker<br /> stably eliminates false particles located on carbon films and ice crystals, independent<br /> of the upstream particle-picking algorithm. By encapsulating contamination<br /> suppression as a universal, model-agnostic post-processing module,<br /> QwenCryoMarker offers a practical, robust, easy-to-deploy toolkit that greatly<br /> improves particle-set quality without modifying existing cryo-EM workflows.<br /> The framework is fully open-source and can be seamlessly integrated into<br /> mainstream cryo-EM processing pipelines.<br /> <br /> == Keywords ==<br /> <br /> == Links ==<br /> <br /> https://journals.iucr.org/paper?wan5006<br /> <br /> == Related software ==<br /> <br /> == Related methods ==<br /> <br /> == Comments ==</div>

WikiSysop