In this presentation, we address the pervasive issue of unmeasured technical heterogeneity in microbiome data. Such heterogeneity, often overlooked, can arise from differential sample processing and study design, leading to potential inaccuracies in results. The existing approaches that address such issues are designed for RMA sequencing and microarray experiments, and thus can not work in microbiome studies when the data is more sparse and over-dispersed. To counter this, we propose the Quantile Thresholding (QuanT) approach, a non-parametric method tailored for the complex distribution of microbial read counts. QuanT adeptly identifies and corrects for unmeasured heterogeneity, thus refining the quality of downstream analytical processes.
We will present our application of QuanT to both synthetic and real microbiome datasets, illustrating its capacity to reveal hidden variabilities and enhance analytical outcomes, especially focusing on the FDR control. The implementation of QuanT promises to fortify the integrity of conclusions drawn from microbiome studies, offering a significant step forward in the field's methodological toolkit.
I | Attachment | Action | Size | Date | Who | Comment |
---|---|---|---|---|---|---|
png | N._Zhao_headshot.png | manage | 307 K | 25 Mar 2024 - 10:30 | CierraStreeter |