HIV Cure Research Center, Ghent University
Abstract internship (advanced bachelor of bioinformatics) 2016-2017: ddpcRquant: RShiny webtool for data analysis of single channel ddPCR experiments
Absolute quantification of nucleic acids by digital PCR (dPCR) is gaining rapid interest in the field of molecular biology and clinical diagnostics because dPCR offers an improved accuracy and sensitivity over quantitative PCR and cancels the need of standard dilution series. These benefits arise from the different methodology used by the dPCR setup. In dPCR, the sample is divided into multiple partitions (3000 - 2 million) with each partition representing an isolated endpoint PCR reaction. After PCR cycling, the fluorescently labeled probes enable the readout of fluorescence built-up and allow the classification of each partition as positive or negative. As a result, the concentration of nucleic acid templates in the sample can be calculated based on Poisson distribution with the number of positive, negative and total partitions. To date, the droplet digital PCR system (Bio-Rad) is the most widely used dPCR platform and the HIV Cure Research Center at Ghent University (internship) has ample experience with optimizing diagnostic HIV assays on this platform.
However, this current generation of dPCR platforms still need careful validation and struggle with the lack of a standardized data analysis method that impedes reproducible quantification results and universal reporting of dPCR data. Therefore, we previously developed a new data analysis pipeline that addresses this problem and offers a much- needed statistical framework (based on extreme value theory) for automated threshold determination and dPCR quantification. In a next phase, a practical tool is highly needed in order for this analytical approach to be adopted by the dPCR community.
In this context, the internship project had the goal to create and provide the dPCR end- user with an easy-to-use and freely available webtool for single channel dPCR experiments that is based on the previously in-house developed statistical pipeline. The application was built with the Shiny package (v1.0.3) that offers a web application framework for the R programming language in order to create user interfaces that communicate with the back- end server via R-based functions (RStudio v1.0.143 with R v3.4.0).
The ddpcRquant webtool runs through a 5-step analysis with corresponding tabs that need to be executed in consequent order. Each of the tabs has a two panel layout with on the left, the control/input panel that changes with each analysis step and on the right, the output panel. The 5 steps include: Data Input/Upload, Overview Assays, Threshold Analysis, Sample Analysis and Summary.
The first two steps of the analysis encompass quality control for the uploaded data. Here, the end-user can check if the correct files were uploaded and if the annotation of the wells correspond to the intended assay and have the appropriate status (negative template control (NTC) or sample).
Next, the basis of the tool is centered around the NTC threshold analysis which calculates the threshold to be used for the sample quantification. The end-user has the choice to use an extreme value theory based threshold determination or a manual threshold. The end- user can switch between these methods during the analysis because Shiny gives the possibility to work with reactive environments - if parameters are changed in the control panel during the NTC threshold analysis, the output can be updated without the need to restart the data upload or entire analysis. In step 4, the threshold is automatically applied to the samples and at the Summary tab, a table overview of the quantification results is given together with the analysis parameters used. All intermediate and final results are available for download and include: NTC threshold plots, sample plots and the summary tables.
To conclude, this tool offers a freely online available analysis pipeline for single channel ddPCR experiments and simplifies analysis for end-users without a bioinformatics background. In addition, there is a built-in demo dataset provided and a how-to-use section to guide the end-user. This Shiny webtool is freely available and does not require a subscription (http://antonov.ugent.be:3838/ddpcrquant/).
De Pintelaan 185