README.md
In lmuenter/coulteR: Make AccuComp-datasets readable in R

coulteR

This package interfaces R with data from the Z2 Coulter Counter. The aim is to make AccuComp-datasets usable in R. First, per-sample datasets are exported into a readable format (.xls) using AccuComp. Next, these datasets are parsed for semantic table boundaries. Lastly, individual tables are extracted into tidy dataframes. For binned measurements of these datasets, {coulteR} enables peak detection with an a priori particle diameter. Weighted mean and standard deviation of target distributions can be used for further analyses.

Install from GitHub:

devtools::install_github("lmuenter/coulteR")

In this demonstration, we will extract the summary module from our AccuComp-tables. Per sample, this module provides an overview over general statistics of the experiment.

# load package
library(coulteR)

# set the path to your files
exp_dir = "data/z2/"

# load all summarise
summaries.df = bulk_read(exp_dir, module = "summary")

We can then extract e.g. the mean particle size for our datasets.

means.df = summaries.df %>%
  filter(var == "Mean") %>%  # retain only mean particle size
  select(-var)               # optional: delete column `var`

In addition to overall sample statistics, AccuComp-datasets contain binned measurements of particle size. If a non-uniform distribution of particle sizes is present, e.g. with xenic cultures or coagulated cells, peak detection can be used to compute parameters of a specific distribution.

1. Extract `measurements`

We start by extracting the measurements-module. It contains binned measurements of diameter, number of cells and a few other parameters of our sample.


# extract the module `measurements` for each dataset
measurements.df = bulk_read(exp_dir, module = "measurements")

# show measurements
measurements.df[c(7:13),]

   bin diameter.bin.um number.diff number.diff.ml volume.diff.perc sample
10   7         3.05711           0              0         0.000000      A
11   8         3.21810           0              0         0.000000      A
12   9         3.36441           2            200         0.016526      A
13  10         3.49952          71           7100         0.655752      A
14  11         3.62444          37           3700         0.377656      A
15  12         3.74130           9            900         0.100618      A
16  13         3.85172           9            900         0.109375      A

Where bin denotes the bin number, diameter.bin.um the diameter of the bin, number.diff the number of particles, number.diff.ml the number of cells per ml. Please note the peak in the number of cells at about 3 um.

Now we will extract and compute distribution parameters of our target peak. In this example, we observed a pronounced peak at around 3 um. To properly characterise the parameters of this tarket distribution of cell sizes, we however need to counteract binning effects. For this, we compute the weighted mean and standard deviation of the particle diameter:


# Approximate diameter of target organisms (here in um)
diameter = 3

# Detect all peaks close to target size
measurements.peaks = bulk_peak_detect(measurements.df, diameter = diameter)

# display result
measurements.peaks

  sample  d.peak d.range.start d.range.end n.cells d.wtdmean    d.wtdsd
1      A 3.49952       3.21810     3.74130     119  3.554376 0.08140310
2      B 3.62444       3.21810     4.14914     697  3.678880 0.18246395
3      C 3.49952       3.21810     3.85172      86  3.563902 0.11228003
4      D 3.11919       2.72458     3.61375     605  3.217490 0.14579105
5      E 3.11919       2.72458     3.52583     107  3.119194 0.12749893
6      F 3.62444       3.36441     3.85172      78  3.598619 0.09493823
7      G 3.23034       2.86862     3.85557     635  3.309034 0.16519370
8      H 3.49952       3.21810     3.85172      56  3.577974 0.10170359

For the target peak of each sample, this dataframe contains:

d.peak peak of the target distribution, binned diameter
d.range.start start of target distribution, binned diameter
d.range.end end of target distribution, binned diameter
n.cells number of cells in target distribution
d.wtdmean weighted mean of particle diameter of the target distribution. Use this for further analysis!
d.wtdsd weighted standard deviaton of particle diameter of the target distribution. Use this for further analysis!

Now we can visualise some tracks with highlighted target peaks. These plots may be useful to determine the success of our extraction method:

ggtrack(measurements.df, measurements.peaks, N = 4, show.legend = FALSE)

When using coulteR::read_accucomp(), other modules can be imported by specifying the parameter module. These are:

|module |content |value --- | --- | --- |all|All modules|A list of dataframes |settings|Experimental setup (device, duration, aperture etc.)|A two-column dataframe |summary|Summary Statistics of the sample (e.g. Mean Particle Size)|A two-column dataframe |sizes_absolute|Size Distribution|A two-column dataframe |sizes_summary|Another Size Distribution|A two-column dataframe |volumes |Cell Volume (fL)|A two-column dataframe |measurements|Measurement Distribution(binned)|A 7-column dataframe