Abstract

Data Analysis of droplet digital PCR with Generalised Linear Mixed Models

Olivier Thas, Professor in Biostatistics & Honorary Professor, Ghent University

Target quantification with droplet digital PCR (ddPCR) depends heavily on the Poisson assumption which makes the data analysis less straightforward than for RT-qPCR. In this talk we demonstrate how a specific class of Generalised Linear Mixed Models (GLMM) can be used for the analysis of ddPCR data. The GLMM statistical modelling framework is very flexible and allows for analysing many experimental designs, including correctly dealing with replicates, run or plate effects and one or more references. The method can be used for absolute quantification, CNV and gene expression, and it allows for standard error and confidence interval calculation and hypothesis testing. Understanding the statistical model may help in improving designs for ddPCR experiments. GLMMs are available in many statistical software. We illustrate the flexibility of the framework using the R software