Research information

• Project Name: Review and Revamp of Microbiome Compositional Data Transformation
• Project director: Prof. Liangliang Zhang
• Project date: Nov 2023 - Present
• Publication: https://doi.org/10.1016/j.csbj.2024.11.003

Conducted a systematic review of compositional data transformations in microbiome data analysis, focusing on the strengths and limitations of existing techniques like rarefaction, scaling, and transformation. Developed a novel framework combining proportion conversion with contrast transformations (CCT Framework) to enhance data handling in high zero-inflation microbiome datasets. Proposed two new transformation methods—Centered Arcsine Contrast (CAC) and Additive Arcsine Contrast (AAC)—which demonstrated improved performance in stabilizing variance and mitigating outlier influence, particularly for high zero-inflation data. These methods support applications such as differential abundance testing, variable selection, and distance-based analyses. Additionally, proposed a dual-group conversion framework for proportional data, including the Dual-Group Truncated Logit (DGTL) conversion, which is beneficial for two-sample testing by allowing transformed data to follow two independent distributions, effectively balancing the preservation of pre-conversion information and enhancing the power to detect differential abundance.