image: Diagram of the MEFE Algorithm and partial result. (a) Elastic feature extraction. (b) MEFE reduces errors by microbiome misclassification or data sparsity. (c) PCoA analysis on Real Dataset I (ASD). (d) Random Forest-based classification on Real Dataset I.
Credit: HIGHER EDUCATON PRESS
Researchers at Qingdao University have developed a novel algorithm, Microbiome Elastic Feature Extraction (MEFE), that significantly improves the identification of microbiome biomarkers by incorporating phylogenetic, taxonomic, and functional relationships among microbes. This advancement addresses longstanding challenges in microbiome research, such as data sparsity and sequencing errors, potentially leading to more accurate disease diagnostics and personalized medicine. The findings were published on 15 January 2026 in Frontiers of Computer Science.
Addressing Challenges in Microbiome Research
Traditional methods for identifying microbiome biomarkers often struggle with high false positive and negative rates due to the complex nature of microbial communities and technical limitations in sequencing. MEFE tackles these issues by elastically integrating the abundance information of a microbe with that of its closely related neighbors, based on evolutionary and functional similarities. This approach allows for a more robust and biologically meaningful identification of biomarkers, enhancing the understanding of microbe-disease associations.
Demonstrated Effectiveness Across Diverse Datasets
The research team evaluated MEFE using both synthetic and real-world 16S rRNA gene sequencing datasets, including samples related to Autism Spectrum Disorder and Type-2 Diabetes. Results showed that MEFE outperformed existing methods in accurately identifying relevant microbial signatures, reducing both false positives and negatives. Analyses such as Principal Coordinate Analysis and Random Forest classification further confirmed MEFE's superior discriminative power.
Implications for Clinical and Ecological Applications
By effectively capturing the complex relationships within microbial communities, MEFE offers a powerful tool for advancing microbiome research. Its ability to provide more accurate biomarker identification holds promise for improving disease diagnostics, informing personalized treatment strategies, and contributing to ecological conservation efforts.
Availability and Further Information
The MEFE algorithm is implemented in Python and is freely available at: https://github.com/qdu-bioinfo/MEFE.
Contact Information
For more details, please contact:
Professor Xiaoquan Su
College of Computer Science and Technology
Qingdao University
Email: suxq@qdu.edu.cn
Journal
Frontiers of Computer Science
Method of Research
Experimental study
Subject of Research
Not applicable
Article Title
MEFE: microbiome signature identification based on elastic feature extraction
Article Publication Date
15-Jan-2026