ggPlantmap is an open-source R package with the goal of facilitating the generation of informative ggplot maps from plant images to explore quantitative cell-type specific data. When combined with external quantitative data, ggPlantmap can be used for the visualization and displaying of spatial profiles in distinct parts/cells of the plant.
Included in the package there is a set of pre-loaded maps created from
previously published plant images that can be directly inserted into a
ggplot coding workflow. ggPlantmap enables users to plot heatmap
signatures of gene expression or any spatial quantitative data onto
plant images providing a customizable and extensible platform for
visualizing, and analyzing spatial quantitative patterns within specific
plant regions This package uses the flexibility of the well-known
ggplot2 R package to allow users to tailor maps to their specific
research questions.
We created a step-by-step user guide to help users navigate through the package.
We also created a step-by-step guide with tips on how to create your ggPlantmap.
Below you can find general instructions on how to navigate through ggPlantmap.
##install devtools (if you haven't already)
install.packages("devtools")
library(devtools)
## Installing from a github respository
install_github("leonardojo/ggPlantmap")Each unique ggPlantmap is a table (tibble) object with points coordinates (x,y) of specific polygons extracted from plant images.
library(ggPlantmap)
head(ggPm.At.roottip.longitudinal)
#> # A tibble: 6 × 7
#> ROI.name Level1 Level2 ROI.id point x y
#> <chr> <chr> <chr> <int> <int> <dbl> <dbl>
#> 1 Meristem.QC Meristem QC 1 1 121. -323.
#> 2 Meristem.QC Meristem QC 1 2 127. -315.
#> 3 Meristem.QC Meristem QC 1 3 134. -315.
#> 4 Meristem.QC Meristem QC 1 4 149. -318.
#> 5 Meristem.QC Meristem QC 1 5 149. -329.
#> 6 Meristem.QC Meristem QC 1 6 134. -327.
The whole list of pre-loaded ggPlantmap objects can be found in the
table ggPm.summary. You can find the description of ggPlantmaps, as well
as information of its creator. Because most ggPlantmaps are based on
previously published plant images, the references for each specific
image can also be found in the summary table. We hope to keep updating
the ggPlantmap catalog, with the help of the plant research community.
head(ggPm.summary)
#> # A tibble: 6 × 9
#> ggPlantmap.name Species Tissue Type Descr…¹ Layers Image…² Made.by Conta…³
#> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr>
#> 1 ggPm.At.roottip.c… Arabid… root cros… Cross-… Cells https:… Leonar… jo.leo…
#> 2 ggPm.At.roottip.l… Arabid… root long… Longit… Cells https:… Leonar… jo.leo…
#> 3 ggPm.At.3weekrose… Arabid… roset… top … Top vi… Leaves https:… Leonar… jo.leo…
#> 4 ggPm.At.leafepide… Arabid… leaf … top … Top vi… Cells https:… Leonar… jo.leo…
#> 5 ggPm.At.leaf.cros… Arabid… leaves cros… Cross-… Cells https:… Leonar… jo.leo…
#> 6 ggPm.At.seed.devs… Arabid… seed deve… Diagra… Cells… https:… Leonar… jo.leo…
#> # … with abbreviated variable names ¹Description, ²Image.Reference,
#> # ³Contact.Info
##Listing all the ggPlantmap objects
ggPm.summary$ggPlantmap.name
#> [1] "ggPm.At.roottip.crosssection"
#> [2] "ggPm.At.roottip.longitudinal"
#> [3] "ggPm.At.3weekrosette.topview"
#> [4] "ggPm.At.leafepidermis.topview"
#> [5] "ggPm.At.leaf.crosssection"
#> [6] "ggPm.At.seed.devseries"
#> [7] "ggPm.At.earlyembryogenesis.devseries"
#> [8] "ggPm.At.shootapex.longitudinal"
#> [9] "ggPm.At.inflorescencestem.crosssection"
#> [10] "ggPm.Sl.root.crosssection"
#> [11] "ggPm.At.leaf.topview"
#> [12] "ggPm.At.rootelong.longitudinal"
#> [13] "ggPm.At.rootmatur.crosssection"
#> [14] "ggPm.At.flower.diagram"
#> [15] "ggPm.At.lateralroot.devseries"
#> [16] "ggPm.Ms.root.crosssection"All ggPlantmaps are pre-loaded in the package, you can call them directly in your R environment by typing their name.
library(ggPlantmap)
##examples
ggPm.At.roottip.longitudinal
#> # A tibble: 1,541 × 7
#> ROI.name Level1 Level2 ROI.id point x y
#> <chr> <chr> <chr> <int> <int> <dbl> <dbl>
#> 1 Meristem.QC Meristem QC 1 1 121. -323.
#> 2 Meristem.QC Meristem QC 1 2 127. -315.
#> 3 Meristem.QC Meristem QC 1 3 134. -315.
#> 4 Meristem.QC Meristem QC 1 4 149. -318.
#> 5 Meristem.QC Meristem QC 1 5 149. -329.
#> 6 Meristem.QC Meristem QC 1 6 134. -327.
#> 7 Meristem.QC Meristem QC 2 1 150. -330.
#> 8 Meristem.QC Meristem QC 2 2 150. -318.
#> 9 Meristem.QC Meristem QC 2 3 156. -317.
#> 10 Meristem.QC Meristem QC 2 4 164. -316.
#> # … with 1,531 more rows
ggPm.At.roottip.crosssection
#> # A tibble: 1,408 × 5
#> ROI.name ROI.id point x y
#> <chr> <int> <int> <dbl> <dbl>
#> 1 Epidermis 1 1 156. -333.
#> 2 Epidermis 1 2 167. -332.
#> 3 Epidermis 1 3 177. -340.
#> 4 Epidermis 1 4 176. -380.
#> 5 Epidermis 1 5 173. -384.
#> 6 Epidermis 1 6 165. -387.
#> 7 Epidermis 1 7 157. -387.
#> 8 Epidermis 1 8 145. -381.
#> 9 Epidermis 1 9 142. -377.
#> 10 Epidermis 1 10 138. -371.
#> # … with 1,398 more rows
ggPm.Ms.root.crosssection
#> # A tibble: 2,441 × 5
#> ROI.name ROI.id point x y
#> <chr> <int> <int> <dbl> <dbl>
#> 1 C1 1 1 270. -308.
#> 2 C1 1 2 234. -287.
#> 3 C1 1 3 241. -257.
#> 4 C1 1 4 271. -238.
#> 5 C1 1 5 285. -243.
#> 6 C1 1 6 307. -270.
#> 7 C1 1 7 298. -289.
#> 8 C1 1 8 284. -302.
#> 9 C1 2 1 285. -242.
#> 10 C1 2 2 308. -270.
#> # … with 2,431 more rowsYou can use the ggPlantmap.plot() function to quickly visualize your ggPlantmap.
##ggPlantmap.plot(data,layer,linewidth=0.5,show.legend=T)
ggPlantmap.plot(ggPm.At.roottip.longitudinal,ROI.id,linewidth = 1,show.legend = F)
If you have experience with ggplot, you can feed your a ggPlantmap object into a ggplot with the geom_polygon() function.
library(ggplot2)
ggplot(ggPm.At.roottip.longitudinal,aes(x,y)) +
geom_polygon(aes(group=ROI.id,fill=factor(ROI.id)),show.legend = F,colour="black",linewidth=1) +
coord_fixed() ## important to keep the aspect ratio of the plot
Because each polygon on ggPlantmap is characterized by specific levels (examples: Region,Stage,Part), you can color map them individually. Using ggPlantmap, you can color map based on unique layers of the ggPlantmap. Just specify the column you want to colormap in the layer option of the function. ggPlantmap.plot() is based on ggplot so you can add specific modifications to it using ggplot coding logic.
library(ggplot2)
library(cowplot)
head(ggPm.At.seed.devseries)
#> # A tibble: 6 × 8
#> ROI.name Stage Part Region ROI.id point x y
#> <chr> <chr> <chr> <chr> <int> <int> <dbl> <dbl>
#> 1 Preglobular.seedcoat.Distal Seed … Preg… seed… Dista… 1 1 277. -693.
#> 2 Preglobular.seedcoat.Distal Seed … Preg… seed… Dista… 1 2 280. -689.
#> 3 Preglobular.seedcoat.Distal Seed … Preg… seed… Dista… 1 3 280. -685.
#> 4 Preglobular.seedcoat.Distal Seed … Preg… seed… Dista… 1 4 285. -681.
#> 5 Preglobular.seedcoat.Distal Seed … Preg… seed… Dista… 1 5 286. -675.
#> 6 Preglobular.seedcoat.Distal Seed … Preg… seed… Dista… 1 6 286. -669.
## Stage: Seed development stage
## Part: Distinct parts of a seed (Seed coat, Endosperm and Embryo)
## Region: Specific regions of each part of the Arabidopsis seed
## Reference: Belmonte, Mark F., et al. "Comprehensive developmental profiles of gene activity in regions and subregions of the Arabidopsis seed." Proceedings of the National Academy of Sciences 110.5 (2013): E435-E444.
a <- ggPlantmap.plot(ggPm.At.seed.devseries,Region,linewidth = 0.5) +
scale_fill_brewer(palette="Set3") +
ggtitle("Regions of Arabidopsis seed development") +
theme(legend.key.height= unit(0.25, 'cm'),
legend.key.width= unit(0.25, 'cm'))
b <- ggPlantmap.plot(ggPm.At.seed.devseries,Stage,linewidth = 0.5) +
scale_fill_brewer(palette="Set1") +
ggtitle("Stages of Arabidopsis seed development") +
theme(legend.key.height= unit(0.25, 'cm'),
legend.key.width= unit(0.25, 'cm'))
c <- ggPlantmap.plot(ggPm.At.seed.devseries,Part,linewidth = 0.5) +
scale_fill_brewer(palette="Set1") +
ggtitle("Parts of Arabidopsis seed development") +
theme(legend.key.height= unit(0.25, 'cm'),
legend.key.width= unit(0.25, 'cm'))
plot_grid(a,b,c,ncol=1,labels=c("a","b","c"),align = "v")
With ggPlantmap you can overlay quantitative data into your ggPlantmap
to visualize it as sort of a heatmap. To do so, you will need another
table with contains quantitative data attributed to your ROIs.
WiYou can combine the ggPlantmap with a external quantitative data using the ggPlantmap.merge() function. later, the heatmap can be generated using the ggPlantmap.heatmap() function.This approach can be very helpful for R Shiny app developers to create web interactive tools to visualize gene expression gene profiles.
##Quantitative sample data, the expression of SCR in distinct cell-types of the Tomato root.
## Adapted data from: Kajala, Kaisa, et al. "Innovation, conservation, and repurposing of gene function in root cell type development." Cell 184.12 (2021): 3333-3348.
head(ggPm.tomatoatlas.expression.sample)
#> # A tibble: 6 × 2
#> Cell.layer SCR.expression
#> <chr> <dbl>
#> 1 Epidermis 1.24
#> 2 Cortex 1.17
#> 3 Endodermis 75.8
#> 4 Phloem 0.44
#> 5 Procambium 0.95
#> 6 Pericycle 0.95
##important: Names in the quantitative data needs to match the ones found in the map.
##Merging both datasets
expression.sample2 <- ggPlantmap.merge(ggPm.Sl.root.crosssection,ggPm.tomatoatlas.expression.sample,id.x = "ROI.name",id.y="Cell.layer") ##Column names between tables are different, need to specify both identifiers in x and y.
head(expression.sample2)
#> # A tibble: 6 × 6
#> ROI.name ROI.id point x y SCR.expression
#> <chr> <int> <int> <dbl> <dbl> <dbl>
#> 1 Exodermis 1 1 615. -370. NA
#> 2 Exodermis 1 2 601. -349. NA
#> 3 Exodermis 1 3 598. -327. NA
#> 4 Exodermis 1 4 617. -312. NA
#> 5 Exodermis 1 5 636. -307. NA
#> 6 Exodermis 1 6 651. -310. NA
##Ploting
ggPlantmap.heatmap(expression.sample2,SCR.expression) +
scale_fill_gradient(low="white",high="Red",na.value ="white")
The principle of creating a ggPlantmap is fairly simple. We generate a list of ROIs (region of interests) in the Icy open-source software (https://icy.bioimageanalysis.org/) from any image. These ROIs are saved as XML files and later be converted into ggPlantmaps by using the function XML.to.ggPlantmap() function. We created step-by-step guide with tips. on how to generate xml images from plant images.
##converting the sample file: ggPm.sample.xml into a ggPlantmap table
ggPm <- XML.to.ggPlantmap("data/ggPm.sample.xml")
head(ggPm)
#> # A tibble: 6 × 5
#> ROI.name ROI.id point x y
#> <chr> <int> <int> <dbl> <dbl>
#> 1 C1 1 1 270. -308.
#> 2 C1 1 2 234. -287.
#> 3 C1 1 3 241. -257.
#> 4 C1 1 4 271. -238.
#> 5 C1 1 5 285. -243.
#> 6 C1 1 6 307. -270.
##plotting the ggPm
ggPlantmap.plot(ggPm)
Not at all. ggPlantmap can also be used to produce many other type of plots. Essentially anything that you can trace, you can create! Be creative! We hope to build a community where people explore the usage of ggPlantmap for the communication of Plant science.
YES!!! Any Plant map can be included in the package. If you create one, please email me ([email protected]) your ggPlantmap as tab-delimited table and I’ll make sure to include in the package. You will be credited and your information will be displayed in the summary file. I really hope this becomes an organic package with the contribution of the plant research community.
I would like to acknowledge Kaisa Kajala, Lisa Oskam, Monica Garcia Gomez, Pierre Gautrat for testing ggPlantmap before launch. I also would like to acknowledge Andres Romanowski for providing some data for the initial tests of ggPlantmap.