EDDA — Power your design


Experimental Design in Differential Abundance analysis

EDDA is a tool for systematic assessment of the impact of experimental design and the statistical test used on the ability to detect differential abundance. EDDA can aid in the design of a range of common experiments such as RNA-seq , ChIP-seq , Nanostring assays, RIP-seq and Metagenomic sequencing, and enables researchers to comprehensively investigate the impact of experimental decisions on the ability to detect differential abundance. More details of EDDA can be found at Luo, Huaien et al. “The Importance of Study Design for Detecting Differentially Abundant Features in High-Throughput Experiments.” Genome Biology 2014;15(12):527.

This web application exposes some of the functionality of EDDA and can be used to answer the following questions:

Differential Abundance Testing

What are the differentially abundant entities in my experiment?

This module is meant to enable users to easily run a panel of statistical tests on any given dataset, to assess the variability of results across statistical tests, compute the intersection and union of these results and correspondingly select a more robust or comprehensive set of calls for downstream analysis. The use-case here is when a user has already generated all their data and would like a limited comparison of results from various DATs.

Performance Evaluation

What is the power of my experiment to detect differentially abundant entities?

The purpose of this module is to allow users to evaluate the relative performance of various statiscal tests based on the characteristics of their experimental setting. Users can adjust the stringency thresholds for the DATs and immediately assess the impact on performance, without rerunning the DATs. The expected use-case for this module is when users have pilot data and would like to do a systematic evaluation of the DATs.

Experimental Design

Do I have enough data and sufficient number of replicates?

This module allows users to specify desired performance targets and the range of experimental choices that are feasible, to identify combinations that can meet the targets as well as the appropriate DATs that can be used to achieve them. Ideally, users in the planning stages of an experiment would use this module to optimize their experimental design.