Vincent de Boer, March 23rd, 2023
🔬 🤖 🥼
This is an import and analysis script written for TRF mitoXpress experiments ran on a SpectraMax. Please have a good look at the data file and key file.
For the script to run the following libraries are needed:
library(tidyverse)
library(broom)
library(here)
library(janitor)
library(lubridate)
library(readxl)
library(bayestestR)
library(pspline)
library(colorspace)
library(ggrepel)
library(ggdist)Also the renv lockfile in this repository is up to date with all packages that are used in this project.
This repository can be cloned to your own github page or the files can be copied to your own Rstudio project.
The data and key files are living in the folder inst\extdata.
A data file starts its data column names on row three. The script skips
the first two lines. The file contains a Time column and a
Temperature column followed by one column for each well. The script
removes rows that contain NaN and also removes the closing rows of the
documents.
The data file is a tab-delimited text file. Apparently our Spectramax spits out txt files in UTF16 unicode format, which is somewhat outdated, since practically all unicode text files now are encoded as UTF8. This is likely the reason why Excel has difficulty reading the exported files from the spectramax properly.
The tab-delimited text column names are also renamed. Please notice that
well names need to be three characters, so A1 becomes A01. This
conversion is also incorporated in the script since the plate reader
outputs two and three digit well names. This is annoying when sorting
data and plotting data.
The imported and processed data looks like this in R:
mito_express_data
#> # A tibble: 610 × 5
#> time time_sec temp well fluorescence
#> <Period> <dbl> <dbl> <chr> <dbl>
#> 1 0S 0 27 C08 42318.
#> 2 0S 0 27 C09 50590.
#> 3 0S 0 27 D08 27063.
#> 4 0S 0 27 D09 27544.
#> 5 0S 0 27 E08 13598.
#> 6 0S 0 27 E09 12513.
#> 7 0S 0 27 F08 9162.
#> 8 0S 0 27 F09 8408.
#> 9 0S 0 27 G08 7690.
#> 10 0S 0 27 G09 8232.
#> # … with 600 more rowsThe key file is needed for the user to manually add meta info to the
analysis. The most important is the choice for time frame for slope
calculation. The time frame is given in seconds (start and end).
Also sample_type and group information should be added.
The key file is an excel file (.xlsx file).
The key file in R looks like:
key_file
#> # A tibble: 10 × 6
#> well sample_type group window start end
#> <chr> <chr> <chr> <chr> <dbl> <dbl>
#> 1 C08 sample GOx 10U window1 0 120
#> 2 C09 sample GOx 10U window1 0 120
#> 3 D08 sample GOx 5U window1 0 120
#> 4 D09 sample GOx 5U window1 0 120
#> 5 E08 sample GOx 1U window2 300 780
#> 6 E09 sample GOx 1U window2 300 780
#> 7 F08 blank no enzyme GOx 0U window2 300 780
#> 8 F09 blank no enzyme GOx 0U window2 300 780
#> 9 G08 blank no glucose GOx 10U no glucose window2 300 780
#> 10 G09 blank no glucose GOx 10U no glucose window2 300 780The first plot is the raw data traces with the smoothed spline.
transformed <- readRDS(here::here("data", "transformed_4dpf_example.rds"))
transformed %>%
ggplot(aes(x=time_sec, y = (fluorescence), color = well, group = group))+
geom_point()+
geom_line(aes(y = (zero_order), color = well, group = well),
alpha = 0.5,
linewidth = 2)+
ggrepel::geom_label_repel(data = . %>% filter(time_sec == max(time_sec)),
aes(label = well),
show.legend = FALSE)+
colorspace::scale_colour_discrete_divergingx(palette = "Geyser", rev = FALSE)+
labs(y = "fluorescence (AU)",
x = "time (sec)") +
facet_wrap(~group, scales = "fixed")+
theme_bw(base_size = 18)
The second plot is the first derivative of the smoothed spline, it demosntrated were the maximum slope is.
transformed %>%
ggplot(aes(x=time_sec, y = first_order, color = well, group = group))+
geom_point()+
ggrepel::geom_label_repel(data = . %>% filter(time_sec == max(time_sec)),
aes(label = well),
show.legend = FALSE)+
colorspace::scale_colour_discrete_divergingx(palette = "Geyser", rev = FALSE)+
labs(y = "change in fluorescence (1st der.)",
x = "time (sec)") +
facet_wrap(~group, scales = "fixed")+
theme_bw(base_size = 18)
This is just a plot of the log transform of the smoothed spline.
Here the slopes are plotted for each well and plotted for each group as individual points and as point_interval (see also ggdist package). The dot is the median of the datapoints, the thick bar is where 95% of the data is and the thin bar is where 80% of the data is.
transformed %>%
filter(!group %in% c("empty")) %>%
slice(1, .by = well) %>%
ggplot(aes(x=group, y = log_slope_till_max, color = well, group = group))+
geom_point(size = 4)+
ggdist::stat_pointinterval(position = position_dodge(width = -0.4, preserve = "single"))+
colorspace::scale_colour_discrete_divergingx(palette = "Geyser", rev = FALSE)+
labs(y = "log slope (AU/1000sec)",
x = "") +
theme_bw(base_size = 18)+
theme(panel.border = element_blank(),
axis.line.y = element_line(),
axis.line.x = element_line())
The data processing goed through several steps. It uses the
smooth.Pspline from the Psline package. The smoothing parameters
(spar = 0.8 and method = 2) are set. The zero-order and first_order
splines are then processed and stored in the transformed tibble. Also,
the area under the curve (AUC) is calculated and the original slope in
the range from time_sec = 800 (s) to time_sec = 2800 (s) is calculated.
These are not used anymore.
To steps that are taken in the prcosssing are presented here:
/Users/vincentdeboer/Documents/R/mitoXpress/mitoXpress_data_analysis_long.png