PoDCall: Positive Droplet Caller for DNA Methylation ddPCR
19 May 2021
Contents
1 Introduction
PoDCall (Positive Droplet Caller) is a package that aims to provide a robust calling of positive droplets in DNA methylation droplet digital PCR (ddPCR) experiments performed on the Bio-Rad platform. PoDCall provides functions that reads files exported from QuantaSoft containing amplitudes from a run of ddPCR (one 96 well plate), sets thresholds for both channels of each individual well and calculates concentrations and normalized concentration for each well. The resulting threshold table can optionally be written to file automatically by the main workflow function. PoDCall also offers functionality for plotting, both individual wells and multiple well plots. Plots for individual wells can be made and saved as .pdf-files as part of the main workflow function, or by calling the various plotting functions individually.
1.1 Gaussian Mixture Models
DdPCR experiments generate a mixture of droplets, positive droplets which contain the target that will be amplified, and negative droplets that does not contain the target and show no amplification. PoDCall relies on fitting Gaussian mixture models to set thresholds for each individual well that will be used to classify the droplets as either positive or negative. For more details on the concepts of PoDCall, see the application note.
1.2 Input Data
The input data is .csv-files exported from ‘QuantaSoft’, and each file contains the amplitude values of droplets from one well of a 96 well plate. The first two columns of the files should have headers ‘Ch1 Amplitude’ and ‘Ch2 Amplitude’. To read in data, use the function importAmplitudeData, which will read all amplitude files in the directory given as argument. Each file will be stored as a named data frame in a list, where the name will be the well ID. For this reason, all raw data files in the directory given as argument should be from the same 96 well plate to avoid well coordinate duplicates.
2 Installation
PoDCall requires some packages to be installed, and if any required packages are not yet installed, the installation of PoDCall should take care of it (you will be prompted to install the packages that are missing).
The released version of PoDCall can not yet be installed from BIOCONDUCTOR, but for now it can be installed from GitHub or from a source file using:
## Install from Bioconductor
if (!requireNamespace("BiocManager", quietly = TRUE))
install.packages("BiocManager")
BiocManager::install("PoDCall")
## Install PoDCall from GitHub
install.packages("devtools")
devtools::install_github("HansPetterBrodal/PoDCall")
## Install PoDCall from source file
install.packages("remotes")
remotes::install_local("path/to/PoDCall_0.99.0.tar.gz")
After installing PoDCall and the required packages, PoDCall can be loaded with:
library(PoDCall)3 Example / Usage
One step of setting thresholds includes a random sampling of droplets to greatly
reduce running time. To ensure reproducible results it is recommended to set a
seed using set.seed().
To run the full PoDCall workflow, call the function podcallDdpcr():
## Run PoDCall
thresholdTable <- podcallDdpcr(dataDirectory="path/to/data/")Where “path/to/data/” is the path of the directory that contains amplitude files from a well plate, in which the files have names that end with "_wellID_amplitude.csv".
3.1 Optional arguments
The following arguments have default values, but can be given other values if desired. For example to write results to file, which is disabled by default.
3.1.1 sampleSheetFile
Path to sample sheet file. Must be a .csv file exported from QuantaSoft and must
include the following columns: Well, Sample, TargetType and Target. The entries
in the column TargetType must be either ‘Ch1Unknown’ or Ch2Unknown, and is used
to extract rows with information from either channel 1 or channel 2. An example
file has been included in the package, which can be found using
system.file("extdata", "Sample_names.csv", package="PoDCall")
3.1.2 B
The number of permutations used by the likelihood ratio test (LRT)
which decides the number of components in the model fitted from the data.
Default value B=200.
3.1.3 Q
A parameter used for calling outliers. Q is multiplied with the interquartile
range of the distribution of amplitude values to determine if droplets of
extreme amplitude values are to be considered outliers. The default value is
Q=9, which has been determined through cell line experiments and testing.
A higher Q will generally result in a higher or more strict threshold. Q
provides an option to adjust how thresholds are set in a systematic and
reproducible way. It is recommended to try a few different values and visually
inspect the results.
3.1.4 refwell
The well used as reference when calculating the shift in baseline
between wells. By default refwell=1, but can be changed in cases where the
first well is not suited to be used.
3.1.5 ch2
If channel 2 is not in use, set ch2 = FALSE to avoid error caused by
empty channel 2 column. Default is ch2=TRUE.
3.1.6 resultsToFile
The user can choose to let PoDCall save the results table as a .csv-file by
setting resultsToFile=TRUE (default: resultsToFile = FALSE). When
resultsToFile is set to TRUE, a results directory will be created where the
result file will be saved. The results directory will have the same name as the
data directory with "_results" added: "path/to/data_results/
3.1.7 plots
The user can choose to make plots that are written to file by setting
plots=TRUE (default: plots=FALSE). Plots will be saved to the results
directory created when resultsToFile=TRUE. The results directory will also be
created if only plots=TRUE.
3.1.8 resPath
Optional argument to specify a directory for writing results file(s) to other
than the results directory created by default. Requires resultsToFile=TRUE.
3.2 Threshold table columns
The table that is returned when running podcall_ddpcr() contains columns with
more or less self-explanatory column names, and well ID (well coordinates) as
rownames:
3.2.1 sample_id
If a sample sheet file is provided, this will have the sample ID from the sample sheet. Otherwise empty
3.2.2 thr_target
the threshold set for channel 1, assumed to be the target
3.2.3 thr_ctrl
The threshold set for channel 2, assumed to be the control
3.2.4 pos_dr_target
The number of positive droplets in channel 1 (target)
3.2.5 pos_dr_ctrl
The number of positive droplets in channel 2 (control)
3.2.6 tot_droplets
Number of droplets.
3.2.7 c_target
Concentration of target, calculated by the formula \(-\log\dfrac{\dfrac{\text{neg_drop_tar}}{\text{tot_droplets}}}{0.000851}\) (where does 0.000851 come from from and what is the name of this parameter) where neg_drop_tar is number of negative droplets in channel 1 (target).
3.2.8 c_ctrl
Concentration of control, calculated by the formula \(-\log\dfrac{\dfrac{\text{neg_drop_ctrl}}{\text{tot_droplets}}}{0.000851}\) where neg_drop_ctrl is number of negative droplets in channel 2 (control).
3.2.9 c_norm
Normalized concentration, calculated by the formula \(\dfrac{\text{c_target}}{\text{c_ctrl}}\cdot400\)
4 PoDCall functions
podcallDdpcr() is the main wrapper function that returns a table with the
results of PoDCall to the user. This function uses a set of functions that read
the amplitude data from file, set thresholds and make plots. All functions
involved can be used individually should the user only want to use some of the
functionality of PoDCall. Also see help files for more details about the
functions and their arguments.
4.1 importAmplitudeData()
Reads .csv-files with amplitude data outputted from QuantaSoft and store the data in a list, one data frame per well. Each element in the list will be named using it’s well ID (coordinate) of the 96 well plate that the sample belong to.
## Path to example data files included in PoDCall
path <- system.file("extdata", "Amplitudes/", package="PoDCall")
## Read in data files
dataList <- importAmplitudeData(dataDirectory=path)
str(dataList)
#> List of 5
#> $ A04:'data.frame': 18739 obs. of 2 variables:
#> ..$ Ch1: num [1:18739] 940 971 971 976 985 ...
#> ..$ Ch2: num [1:18739] 11795 7868 8377 10007 9523 ...
#> $ B04:'data.frame': 16933 obs. of 2 variables:
#> ..$ Ch1: num [1:16933] 980 995 1002 1007 1014 ...
#> ..$ Ch2: num [1:16933] 9524 7999 7686 7799 9510 ...
#> $ D04:'data.frame': 11713 obs. of 2 variables:
#> ..$ Ch1: num [1:11713] 1042 1070 1094 1112 1112 ...
#> ..$ Ch2: num [1:11713] 7826 9934 7698 7605 7743 ...
#> $ D05:'data.frame': 12642 obs. of 2 variables:
#> ..$ Ch1: num [1:12642] 1045 1057 1063 1068 1079 ...
#> ..$ Ch2: num [1:12642] 9722 7752 9103 7716 7738 ...
#> $ H05:'data.frame': 19638 obs. of 2 variables:
#> ..$ Ch1: num [1:19638] 1043 1094 1098 1104 1119 ...
#> ..$ Ch2: num [1:19638] 7231 7063 7161 6863 7416 ...4.2 importSampleSheet()
Reads a .csv-file outputted from QuantaSoft to get information about the samples: Sample name/id, Assay for target and control.
## Path to example files included in PoDCall
path <- system.file("extdata", "Sample_names.csv", package="PoDCall")
## Select wells to get information for
well_id <- c("A04", "B04", "D04")
## Read in sample sheet information for selected wells
sampleSheet <- importSampleSheet(sampleSheet=path, well_id=well_id)
print(sampleSheet)
#> well_id sample_id target_assay ctrl_assay
#> 1 A04 SW1463 VIM new4Plex
#> 2 B04 SW403 VIM new4Plex
#> 3 D04 SW480 VIM new4Plex4.3 podcallThresholds()
Takes a list of data frames, one for each well, as argument and sets
individual thresholds for each channel of each well. It returns a table with
thresholds, number of positive droplets, concentrations etc. The number of
permutations for likelihood ratio test is by default set to B=400 as a
compromise between run time and stability of the results. The parameter for
calling outliers is by default set to Q=9. Higher Q means more conservative
(higher) thresholds, lower Q will result in over all lower thresholds.
## Path to example data files included in PoDCall
path <- system.file("extdata", "Amplitudes/", package="PoDCall")
## Read in data files
ampData <- importAmplitudeData(dataDirectory=path)
## Calculate thresholds, metrics, concentrations
thresholdTable <- podcallThresholds(plateData=ampData)
print(thresholdTable)4.4 podcallChannelPlot()
Takes the threshold and amplitude values corresponding to a channel of a well as arguments, calls functions that makes scatter plot and histogram and draws a plot with both.
## Read in data and threshold table
path <- system.file("extdata", "Amplitudes/", package="PoDCall")
ampData <- importAmplitudeData(dataDirectory=path)
data("thrTable")
thresholdTable <- thrTable
## Select channel and well to plot
ch <- 1 # target channel
well_id <- names(ampData)[1] # First well in list
## Plot title
plotTitle <- paste0(well_id, ", Ch", ch)
## Create plot
podcallChannelPlot(channelData=ampData[[well_id]][,ch],
thr=thresholdTable[well_id, "thr_target"],
channel=ch,
plotId=plotTitle)
4.5 podcallScatterplot()
Takes the threshold and amplitude values corresponding to a channel of a well as argument and returns a scatter plot.
## Read in data and threshold table
path <- system.file("extdata", "Amplitudes/", package="PoDCall")
ampData <- importAmplitudeData(dataDirectory=path)
thresholdTable <- thrTable
## Select channel and well to plot
ch <- 1 # target channel
well_id <- names(ampData)[1] # First well in list
## Plot title
plotTitle <- paste0(well_id, ", Ch", ch)
## Create plot
podcallScatterplot(channelData=ampData[[well_id]][,ch],
thr=thresholdTable[well_id, "thr_target"],
channel=ch,
plotId=plotTitle)
4.6 podcallHistogram()
Takes the threshold and amplitude values corresponding to a channel of a well as argument, and returns a histogram.
## Read in data and threshold table
path <- system.file("extdata", "Amplitudes/", package="PoDCall")
ampData <- importAmplitudeData(dataDirectory=path)
thresholdTable <- thrTable
## Select channel and well to plot
ch <- 1 # target channel
well_id <- names(ampData)[1] # First well in list
## Plot title
plotTitle <- paste0(well_id, ", Ch", ch)
## Create plot
podcallHistogram(channelData=ampData[[well_id]][,ch],
thr=thresholdTable[well_id, "thr_target"],
channel=ch,
plotId=plotTitle)
4.7 podcallMultiplot()
takes a list of data frames with amplitude data, one per well, and their respective thresholds as arguments and returns faceted scatter plots suitable for comparing wells.
## Read in data and threshold table
path <- system.file("extdata", "Amplitudes/", package="PoDCall")
ampData <- importAmplitudeData(dataDirectory=path)
thresholdTable <- thrTable
## Channel to plot
ch <- 1
## Create comparison plot
podcallMultiplot(plateData=ampData,
thresholds=thresholdTable[names(ampData),],
channel=ch)
5 PoDCall shiny application
PoDCall does also include an application powered by shiny that launches in a web browser. The application provides a user friendly and interactive interface to the functionality of PoDCall. To start the app:
podcallShiny()6 PodCall example data
There are some amplitude files and a sample sheet included in the package that are intended to be used to run examples and to try out the functionality of PoDCall. The data files are from a real experiment performed with cell line samples. There is also a threshold table computed from the example data included. PoDCall takes a few minutes to run due to bootstrapping, and this table is used in examples for functions where threshold is an argument.
6.1 Cell Line Amplitude Data
The cell line amplitude data files can be found in the “extdata” subdirectory
of the package directory and can be found using system.file():
## Path to files
path <- system.file("extdata", "Amplitudes/", package="PoDCall")
## List files
list.files(path)
#> [1] "VIM_4Plex_A04_Amplitude.csv" "VIM_4Plex_B04_Amplitude.csv"
#> [3] "VIM_4Plex_D04_Amplitude.csv" "VIM_4Plex_D05_Amplitude.csv"
#> [5] "VIM_4Plex_H05_Amplitude.csv"The control assay used for the samples in the example data files is an assay developed in-house called 4Plex H. Pharo et al.
6.2 Calculated Threshold Table
The already calculated threshold table is instantly available when PoDCall is
loaded, and is available as an object called thrTable. See ?thrTable for
help file with documentation on the table.
## The threshold table
thrTable
#> sample_id thr_target thr_ctrl pos_dr_target pos_dr_ctrl tot_droplets
#> A04 SW1463 2761 9157 2479 12887 18739
#> B04 SW403 2739 8661 660 7459 16933
#> D04 SW480 2863 8519 44 8336 11713
#> D05 IVDZ_bisulf 2818 8502 1675 6573 12642
#> H05 NTC 2823 7940 0 0 19638
#> c_target c_ctrl c_norm q target_assay ctrl_assay
#> A04 166.700 1368.0 48.74 9 VIM new4Plex
#> B04 46.720 682.4 27.39 9 VIM new4Plex
#> D04 4.423 1461.0 1.211 9 VIM new4Plex
#> D05 167.000 862.3 77.47 9 VIM new4Plex
#> H05 0.000 0.0 No DNA 9 VIM new4Plex7 Session info
Here is the output of sessionInfo() on the system on which this document was
compiled
sessionInfo()
#> R version 4.1.0 (2021-05-18)
#> Platform: x86_64-pc-linux-gnu (64-bit)
#> Running under: Ubuntu 20.04.2 LTS
#>
#> Matrix products: default
#> BLAS: /home/biocbuild/bbs-3.13-bioc/R/lib/libRblas.so
#> LAPACK: /home/biocbuild/bbs-3.13-bioc/R/lib/libRlapack.so
#>
#> locale:
#> [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
#> [3] LC_TIME=en_GB LC_COLLATE=C
#> [5] LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
#> [7] LC_PAPER=en_US.UTF-8 LC_NAME=C
#> [9] LC_ADDRESS=C LC_TELEPHONE=C
#> [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
#>
#> attached base packages:
#> [1] stats graphics grDevices utils datasets methods base
#>
#> other attached packages:
#> [1] PoDCall_1.0.0 BiocStyle_2.20.0
#>
#> loaded via a namespace (and not attached):
#> [1] tidyselect_1.1.1 xfun_0.23 bslib_0.2.5.1
#> [4] shinyjs_2.0.0 purrr_0.3.4 colorspace_2.0-1
#> [7] vctrs_0.3.8 generics_0.1.0 htmltools_0.5.1.1
#> [10] yaml_2.2.1 utf8_1.2.1 rlang_0.4.11
#> [13] jquerylib_0.1.4 pillar_1.6.1 later_1.2.0
#> [16] glue_1.4.2 DBI_1.1.1 lifecycle_1.0.0
#> [19] stringr_1.4.0 munsell_0.5.0 gtable_0.3.0
#> [22] htmlwidgets_1.5.3 LaplacesDemon_16.1.4 evaluate_0.14
#> [25] labeling_0.4.2 knitr_1.33 fastmap_1.1.0
#> [28] ps_1.6.0 httpuv_1.6.1 parallel_4.1.0
#> [31] fansi_0.4.2 highr_0.9 Rcpp_1.0.6
#> [34] xtable_1.8-4 readr_1.4.0 scales_1.1.1
#> [37] promises_1.2.0.1 DT_0.18 diptest_0.76-0
#> [40] BiocManager_1.30.15 magick_2.7.2 jsonlite_1.7.2
#> [43] farver_2.1.0 mime_0.10 gridExtra_2.3
#> [46] ggplot2_3.3.3 hms_1.1.0 digest_0.6.27
#> [49] stringi_1.6.2 rlist_0.4.6.1 bookdown_0.22
#> [52] dplyr_1.0.6 shiny_1.6.0 grid_4.1.0
#> [55] cli_2.5.0 tools_4.1.0 magrittr_2.0.1
#> [58] sass_0.4.0 tibble_3.1.2 crayon_1.4.1
#> [61] pkgconfig_2.0.3 ellipsis_0.3.2 data.table_1.14.0
#> [64] rstudioapi_0.13 assertthat_0.2.1 rmarkdown_2.8
#> [67] R6_2.5.0 mclust_5.4.7 compiler_4.1.0