Dan MacLean 06 October, 2022
besthr is a package that creates plots showing scored HR experiments and plots of distribution of means of ranks of HR score from bootstrapping.
You can install from CRAN in the usual way.
install.packages("besthr")
# or for the dev version
#install.packages("devtools")
devtools::install_github("TeamMacLean/besthr")Please cite as
Dan MacLean. (2019). TeamMacLean/besthr: Initial Release (0.3.0). Zenodo. https://doi.org/10.5281/zenodo.3374507
With a data frame or similar object, use the estimate() function to
get the bootstrap estimates of the ranked data.
estimate() has a basic function call as follows:
estimate(data, score_column_name, group_column_name, control = control_group_name)
The first argument after the
library(besthr)
hr_data_1_file <- system.file("extdata", "example-data-1.csv", package = "besthr")
hr_data_1 <- readr::read_csv(hr_data_1_file)## Rows: 20 Columns: 2
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (1): group
## dbl (1): score
##
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
head(hr_data_1)## # A tibble: 6 × 2
## score group
## <dbl> <chr>
## 1 10 A
## 2 9 A
## 3 10 A
## 4 10 A
## 5 8 A
## 6 8 A
hr_est_1 <- estimate(hr_data_1, score, group, control = "A")
hr_est_1## besthr (HR Rank Score Analysis with Bootstrap Estimation)
## =========================================================
##
## Control: A
##
## Unpaired mean rank difference of A (14.9, n=10) minus B (6.1, n=10)
## 8.8
## Confidence Intervals (0.025, 0.975)
## 4.16625, 7.82625
##
## 100 bootstrap resamples.
plot(hr_est_1)## Picking joint bandwidth of 0.376
You may select the group to set as the common reference control with
control.
estimate(hr_data_1, score, group, control = "B" ) %>%
plot()## Picking joint bandwidth of 0.383
You may select the number of iterations of the bootstrap to perform with
nits and the quantiles for the confidence interval with low and
high.
estimate(hr_data_1, score, group, control = "A", nits = 1000, low = 0.4, high = 0.6) %>%
plot()## Picking joint bandwidth of 0.261
You can extend the estimate() options to specify a third column in the
data that contains technical replicate information, add the technical
replicate column name after the sample column. Technical replicates are
automatically merged using the mean() function before ranking.
hr_data_3_file <- system.file("extdata", "example-data-3.csv", package = "besthr")
hr_data_3 <- readr::read_csv(hr_data_3_file)## Rows: 36 Columns: 3
## ── Column specification ────────────────────────────────────────────────────────
## Delimiter: ","
## chr (1): sample
## dbl (2): score, rep
##
## ℹ Use `spec()` to retrieve the full column specification for this data.
## ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.
head(hr_data_3)## # A tibble: 6 × 3
## score sample rep
## <dbl> <chr> <dbl>
## 1 8 A 1
## 2 9 A 1
## 3 8 A 1
## 4 10 A 1
## 5 8 A 2
## 6 8 A 2
hr_est_3 <- estimate(hr_data_3, score, sample, rep, control = "A")
hr_est_3## besthr (HR Rank Score Analysis with Bootstrap Estimation)
## =========================================================
##
## Control: A
##
## Unpaired mean rank difference of A (5, n=3) minus B (2, n=3)
## 3
## Confidence Intervals (0.025, 0.975)
## 1.33333333333333, 2.66666666666667
##
## Unpaired mean rank difference of A (5, n=3) minus C (8, n=3)
## -3
## Confidence Intervals (0.025, 0.975)
## 7, 9
##
## 100 bootstrap resamples.
plot(hr_est_3)## Picking joint bandwidth of 0.163
In the case where you have use technical replicates and want to see
those plotted you can use an extra plot option which. Set which to
just_data if you wish the left panel of the plot to show all data
without ranking. This will only work if you have technical replicates.
hr_est_3 %>%
plot(which = "just_data")## Picking joint bandwidth of 0.163
You can style plots to your own taste. The object returned from plot()
is a patchwork https://patchwork.data-imaginist.com/ object that
composes two separate plots, the dot plot and the bootstrap percentile
plot, which are themselves ggplot objects. So you can use a mixture of
patchwork annotations functions for whole plot labels and ggplot
themes for individual elements.
You can use the patchwork plot_annotation() function to add titles
library(patchwork)
p <- plot(hr_est_1)
p + plot_annotation(title = 'A stylish besthr plot',
subtitle = "better than ever",
caption = 'Though this example is not meaningful')## Picking joint bandwidth of 0.376
p## Picking joint bandwidth of 0.376
You can change the style of the individual plot elements using
subsetting syntax [[]] . The dot plot can be addressed within the
patchwork object using index 1 within the patchwork object p[[1]],
and the percentile plot using p[[2]]. You must add to the existing
subplot then assign the result back to see the difference in the plot.
Here’s an example that uses theme() to restyle the y-axis text of the
dot plot
library(ggplot2)
p[[1]] <- p[[1]] + theme(axis.title.y = element_text(family = "Times", colour="blue", size=24))
p## Picking joint bandwidth of 0.376
You can change the colours used by the scales in the same way using the
scale functions, though as the type of scale is different for the dot
plot and bootstrap plot you will need to apply a different scale for
each.
For the dot plot, use a discrete scale e.g scale_colour_manual(),
scale_colour_viridis_d() or scale_colour_brewer(type = "qual")
p[[1]] <- p[[1]] + scale_colour_manual(values = c("blue", "#440000"))
p## Picking joint bandwidth of 0.376
p[[1]] <- p[[1]] + scale_colour_viridis_d()## Scale for 'colour' is already present. Adding another scale for 'colour',
## which will replace the existing scale.
p## Picking joint bandwidth of 0.376
p[[1]] <- p[[1]] + scale_colour_brewer(type="qual", palette="Accent")## Scale for 'colour' is already present. Adding another scale for 'colour',
## which will replace the existing scale.
p## Picking joint bandwidth of 0.376
For the percentile plot, use only scale_colour_manual() with specified
colours. Annoyingly, this rewrites the other values associated with the
scale each time, so you’ll need to replace those.
p[[2]] <- p[[2]] + scale_fill_manual(
values = c("blue", "pink", "yellow"),
name = "bootstrap percentile", labels=c("lower", "non-significant", "higher"),
guide = guide_legend(reverse=TRUE)
)## Scale for 'fill' is already present. Adding another scale for 'fill', which
## will replace the existing scale.
p## Picking joint bandwidth of 0.376
