Yiqian Zhang

Z. Zhu, Y. Zhang, S. Lin, and L. Zhang. "DeSHAPE: Decomposing Ecological Structure through Heterogeneity, Asymmetry and Pattern Evaluation of α-Diversity." Bioinformatics Advances.

Z. Zhu, Y. Zhang, R. Liu, W. Li, Z. Huang, X. Chen, Y. Duan, and L. Zhang*. "Koinetic: A Graph-based Distance as a New Ecological Metric for Microbial Community Differential Analysis." Briefings in Bioinformatics.

Z. Zhu, Y. Zhang, W. Li, M. Greenacre, S. Saha, Y. Shi, and L. Zhang. "Mathematical Foundations of Beta Diversity: Why Common Metrics Fail in Microbiome Analysis." Journal of Statistical Theory and Applications, vol. 25, article 9.

Abstract

Beta-diversity analysis is central to comparing microbiome communities, yet widely used dissimilarity measures do not always satisfy the metric, Euclidean, or conditionally negative definite properties required by common downstream methods. We systematically evaluate popular measures, organize them into four mathematical classes, and show how property violations can distort PCoA, PERMANOVA, and kernel-based inference. We also introduce diagnostic tools and correction strategies that restore valid geometry while preserving ordination structure. Analyses across real microbiome datasets translate the theory into practical guidance, supported by an R package, interactive Shiny application, and reproducible tutorials for selecting and refining beta-diversity metrics.

Y. Zhang, J. Schluter, L. Zhang, X. Cao, R. R. Jenq, H. Feng, J. Haines, and L. Zhang. "Review and Revamp of Compositional Data Transformation: A New Framework Combining Proportion Conversion and Contrast Transformation." Computational and Structural Biotechnology Journal, vol. 23, pp. 4088-4107.

Abstract

Microbiome abundance data combine unequal sequencing depth, compositional constraints, and frequent zeros, which can make standard scaling and transformation methods yield inconsistent conclusions. This work reviews the connections and distinctions among existing approaches and introduces a unified framework that pairs proportion conversion with contrast transformation. The framework recovers familiar methods such as ALR and CLR while enabling new transformations, including Centered Arcsine Contrast and Additive Arcsine Contrast. Simulation and data analyses indicate that the arcsine-based methods are particularly useful under heavy zero inflation, whereas log-ratio methods perform better when zeros are less prevalent, offering practical guidance for robust microbiome analysis.