ddPCRclust: an R package and Shiny app for automated analysis of multiplexed ddPCR data

Abstract

MotivationDroplet digital PCR (ddPCR) is an emerging technology for quantifying DNA. By partitioning the target DNA into ∼20 000 droplets, each serving as its own PCR reaction compartment, a very high sensitivity of DNA quantification can be achieved. However, manual analysis of the data is time consuming and algorithms for automated analysis of non-orthogonal, multiplexed ddPCR data are unavailable, presenting a major bottleneck for the advancement of ddPCR transitioning from low-throughput to high-throughput.ResultsddPCRclust is an R package for automated analysis of data from Bio-Rad’s droplet digital PCR systems (QX100 and QX200). It can automatically analyze and visualize multiplexed ddPCR experiments with up to four targets per reaction. Results are on par with manual analysis, but only take minutes to compute instead of hours. The accompanying Shiny app ddPCRvis provides easy access to the functionalities of ddPCRclust through a web-browser based GUI.Availability and implementationR package: https://github.com/bgbrink/ddPCRclust; Interface: https://github.com/bgbrink/ddPCRvis/; Web: https://bibiserv.cebitec.uni-bielefeld.de/ddPCRvis/.Supplementary information Supplementary data are available at Bioinformatics online.

Machine learning algorithm improved automated droplet classification of ddPCR for detection of BRAF V600E in paraffin-embedded samples

“…Of note, the automatic analysis of the droplets failed with ddPCRclust package 6 and ddPCR package 7 for all our samples (results not shown). On the other hand, the k-Nearest Neighbors (k-NN) algorithm in twoddPCR package, which is based on lazy machine learning, was the only one that successfully classified the droplets.…”

“…One example that needs improvement is to overcome the droplet allocation bias in ddPCR analysis. However, most algorithms used to automatically classify droplets were tested using serial dilution of one sample and did not compare to other techniques that are also sensitive such as pyrosequencing 6 – 8 . To fill this gap, we sought to compare classification algorithms with the most reliable method that we found, which was pyrosequencing.…”

“…For automatic detection of ddPCR, three packages that can analyze and visualize ddPCR data in R and are free to use and available online were used ddPCRclust package 6 , ddPCR package 7 , and twoddPCR package 8 .…”

Collaborative Cross Founder Expression Analysis (CCFEA)

“…Shiny is a robust platform frequently used in the biomedical sciences. (Brink et al 2018, Zhang et al 2018).…”

Computational challenges in detection of cancer using cell-free DNA methylation

“…ddPCR is one of the most frequently used techniques these days with multiplex quantification. Various automated algorithms have been developed for ddPCR data analysis namely ‘definetherain’ [57] , ‘ddpcRquant’ [58] , ‘ddpcr’ [59] , ‘twoddpcr’ [60] , ‘ddPCRclust’ [61] , ‘ddPCRmulti’ [62] etc. According to Dobnik et al, [61] the data analysis of such multiplex assays becomes difficult and noisy due to several possible target combinations along with probes cross hybridization in a single droplet.…”