therapeutic-effects-acorus-calamus-l-extract-on-radiation-induced

Therapeutic Effects of Acorus calamus L. Extract on Radiation-induced Skin Injury in a Rat Model

Radiation-induced skin injury is a common and debilitating complication in cancer therapy, often resulting in delayed wound healing and increased patient ...

Northeastern University

Cryo-electron microscopy (cryo-EM) is a powerful tool for high-resolution structural analysis of proteins and viruses. However, a major challenge in cryo-EM data processing is the presence of heterogeneous samples, IC contamination, and extraneous impurities, which hinder accurate target protein identification. To address this issue, we propose the Universal Model for Cryo-electron Microscopy Particle Picking (UM-CPP), a novel framework that integrates feature engineering with deep learning to enhance particle detection in cryo-EM micrographs. The key contribution of UM-CPP lies in its hybrid approach, which combines classical machine learning features with state-of-the-art deep learning techniques. This fusion enables robust and adaptable performance across diverse protein structures while maintaining high accuracy. In comparative evaluations, UM-CPP outperforms existing deep-learning-based methods in detection precision. Additionally, our model provides interpretable feature analysis, offering researchers deeper insights into the decision-making process of particle selectiona critical advancement for improving trust and usability in cryo-EM data analysis. By improving both accuracy and interpretability, UM-CPP advances the field of cryo-EM, facilitating more reliable and efficient structural studies of biological macromolecules.

UM-CPP: A Universal Model for Efficient Classification of Protein Particles in cryo-EM Micrographs with Feature Engineering.

N7-methylguanosine (m7G) is a common RNA modification linked to multiple diseases. Accurate detection of m7G sites is vital for elucidating its biological roles, but conventional methods are often laborious and costly. AI-based approaches offer alternatives, yet most rely on balanced datasets, ignoring real-world imbalances where negative samples vastly outnumber positives. This discrepancy may lead to overestimated model performance. To address this, we reconstructed independent test sets with low positive-to-negative ratios and benchmarked various models. We propose OCm7G, an ensemble of one-class classifiers with hierarchical thresholding. OCm7G achieves performance comparable to state-of-the-art methods on balanced sets and surpasses them in highly imbalanced scenarios, despite using only 52.5 % of the training data. Moreover, OCm7G offers interpretable predictions, aiding researchers in understanding model decisions. The source code and datasets are publicly available at: https://github.com/lidaosheng/OCm7G.

Article	Year
UM-CPP: A Universal Model for Efficient Classification of Protein Particles in cryo-EM Micrographs with Feature Engineering. ACS omega\| PubMed ID: 40686975	2025
OCm7G: An interpretable one-class predictor for m7G methylation sites trained with limit negative samples. Genomics\| PubMed ID: 41389903	2025

Wenxuan Luo

Biography

JoVE Journal Publications

Other Publications