https://github.com/bradyajohnston/reluxr
Deconvolute Luminescence Readings on Bacterial Culture Plates
https://github.com/bradyajohnston/reluxr
plate-reader r r-package synthetic-biology
Last synced: 2 months ago
JSON representation
Deconvolute Luminescence Readings on Bacterial Culture Plates
- Host: GitHub
- URL: https://github.com/bradyajohnston/reluxr
- Owner: BradyAJohnston
- License: other
- Created: 2022-02-24T09:43:11.000Z (about 3 years ago)
- Default Branch: main
- Last Pushed: 2023-03-20T12:39:30.000Z (about 2 years ago)
- Last Synced: 2024-12-31T18:34:38.374Z (4 months ago)
- Topics: plate-reader, r, r-package, synthetic-biology
- Language: R
- Homepage: https://bradyajohnston.github.io/reluxr/
- Size: 7.18 MB
- Stars: 2
- Watchers: 2
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.Rmd
- Contributing: .github/CONTRIBUTING.md
- License: LICENSE
- Code of conduct: .github/CODE_OF_CONDUCT.md
- Support: .github/SUPPORT.md
Awesome Lists containing this project
README
---
output: github_document
bibliography: '`r system.file("REFERENCES.bib", package="reluxr")`'
---
```{r, include = FALSE}
knitr::opts_chunk$set(
collapse = TRUE,
comment = "#>",
fig.path = "man/figures/README-",
out.width = "100%"
)
```
# reluxr
[](https://github.com/BradyAJohnston/reluxr/actions/workflows/R-CMD-check.yaml) [](https://CRAN.R-project.org/package=reluxr) [](https://bradyajohnston.r-universe.dev) [](https://bradyajohnston.r-universe.dev)
Implements the deconvolution algorithm developed in Mauri, Vecchione, and Fritz (2019) which enables deconvolution of luminscence readings for experimental culture plates.
{reluxr} provides functions for calculating the 'best' deconvolution matrix from a calibration plate, and enables usage of this calibration matrix (or one calculated previously) to adjust luminescent experimental values from a plate reader.
This is an R-based implementation of the MatLab workflow from the original paper paper titled *1Deconvolution of Luminescence Cross-Talk in High-Throughput Gene Expression Profiling'* [@mauri2019]
## Installation
The package is not currently available on CRAN.
Install the released version from `r-universe` with the following code:
``` r
install.package("reluxr", repos = "https://bradyajohnston.r-universe.dev")
```
## Example
This is a basic example which shows you how to solve a common problem:
```{r}
#| label: setup
#| message: false
#| warning: false
library(reluxr)
fl <- system.file(
"extdata",
"calibrate_tecan",
"calTecan1.xlsx",
package = "reluxr"
)
```
## Create a Deconvolution Matrix
The deconvolution matrix is created from a calibration plate, which contains a single well with luminescent bacteria, with all other wells being empty.
The cross-talk when the plate-reader measures the wells can then be calculated and used to create a deconvolution matrix which will remove the crosstalk from the measured values.
```{r}
dat <- plate_read_tecan(fl)
dat
```
Regardless of how you read in the required data, it needs to be (and should be regardless) in a _tidy_ format, with each row being an observation and each column a variable. In the case above we have columns for the `cycle_nr`, `time_s`, `signal`, `well` and `value`. While it would be better to have a column for the OD600 and for the LUMI data, the time points do not match and aren't currently pivotable.
To have a look at the final data, we can plot the plate based on log-transformed luminescence values. We can see the very bright single well than contains the bacteria, and the bleed-through signal that is around it.
```{r}
dat_fil <- dat |>
dplyr::filter(signal == "LUMI", time_s > 500)
dat_fil |>
dplyr::group_by(well) |>
dplyr::summarise(value = mean(value)) |>
rl_plot_plate(value)
```
We can use the `rl_calc_decon_matrix()` function which will calculate a deconvolution matrix, reducing the background bleed-through from the plate to below a noise threshold. The noise threshold should be the instrument's background noise, which is defined as three times the standard deviation of a blank well. The lower the noise threshold, the harder it will be to calculate a deconvolution matrix which works with the data.
We can also quickly look at the resulting deconvolution matrix (`mat_d_best`) itself.
```{r}
mat_d_best <- rl_calc_decon_matrix(
data = dat_fil,
value = value,
time = time_s,
ref_well = "E05",
b_noise = 20
)
image(log10(mat_d_best))
```
## Deconvoluting the Data
Now that we have the matrix, we can use it to adjust the data.
We do so using the `rl_adjust_plate()` funciton, which takes a dataframe, the name of the column you which to adjust, the deconvolution matrix (in this case `mat_d_best`, and the name ofthe column which stores the time data).
The returned dataframe will have the value column adjusted and deconvoluted using the deconvolution matrix supplied.
In the examples below we first plot all of the values without deconvolution, then apply the deconvolution matrix and plot the values again.
```{r}
#| code-fold: true
rl_plot_time <- function(data, time, value, group = "well") {
data <- dplyr::mutate(
data,
time = {{ time }},
value = {{ value }},
group = {{ group }}
)
plt <- ggplot2::ggplot(
data,
mapping = ggplot2::aes(
x = time,
y = value,
group = group
)
) +
ggplot2::geom_line() +
ggplot2::scale_y_log10() +
ggplot2::theme_bw()
plt
}
```
#### Raw Values
```{r}
dat |>
dplyr::filter(signal == "LUMI") |>
rl_plot_time(time_s, value, well) +
ggplot2::labs(
x = "Time (s)",
y = "LUM"
)
```
#### Deconvoluted Values
```{r}
dat |>
dplyr::filter(signal == "LUMI") |>
rl_adjust_plate(value, mat_d_best, time = time_s) |> # deconvolute the values
rl_plot_time(time_s, value, well) +
ggplot2::labs(
x = "Time (s)",
y = "LUM"
)
```
We can also replot the plate from earlier, with the newly deconvoluted values.
```{r}
dat_fil |>
rl_adjust_plate(value, mat_d_best, time = time_s) |> # deconvolute the values
dplyr::group_by(well) |>
dplyr::summarise(value = mean(value)) |>
rl_plot_plate(value) +
ggplot2::scale_fill_viridis_c(
"log10(LUMI)",
breaks = 1:5,
limits = c(1, NA)
)
```