toxpi / toxpir

toxpiR R package for the Toxicological Priority Index (ToxPi) algorithm.

toxpiR

R package for the Toxicological Priority Index (ToxPi) prioritization algorithm. Package developed and maintained by the Reif Lab.

Installation

Current stable release (CRAN):

install.packages("toxpiR")

Current stable release (Build from GitHub):

remotes::install_github("ToxPi/toxpiR", 
                        dependencies = TRUE)
                        
Note: Users may need to ensure "remotes" package and packages
requiring "BiocManager" are installed before building package.

if (!require(remotes)) install.packages("remotes")

if (!require(BiocManager, quietly = TRUE)) {
  install.packages("BiocManager")
}
BiocManager::install(c("S4Vectors","BiocGenerics"))

Current stable release (Build from GitHub with vignettes):

remotes::install_github("ToxPi/toxpiR",
                        dependencies = TRUE, 
                        build_vignettes = TRUE)

Note: Building packages with vignettes requires the package
"pandoc" to be installed.

toxpir's Issues

Add tests for txpExportGui

txpExportGui function now in dev branch as of filing this issue. Still needs test coverage.

Documentation request

Terrific work to create this new package.
One observation I encountered when performing the install - and might just be specific to me but figured I would report it nonetheless. Using the install.packages and setting dependencies to True still failed to install S4Vectors and BiocGenerics.
I found I needed to install 'BiocManager' and then use this to install the 2 dependencies. Steps I went through were as follows:
install.packages('BiocManager')
BiocManager::install(c('S4Vectors', 'BiocGenerics'))
install.packages('toxpiR')

Legend issues

Hello,

I was wondering if there is a native way to control legend positions/text/colors?

The default seems to be having some trouble in my case:

As you can see, the slice names are a bit long - and, not all the slices appear to be assigned a color. Any advice would be helpful!

Thanks,
Matt

Issue-19: Changing project maintainer

Update project/documentation for switching toxpiR maintainer to Jonathon Fleming as discussed and prep for new CRAN version

Create donutGrob

Create donutGrob and associated functions for plotting with missing-ness.

Add function for exporting to GUI format

The difficult aspect of doing this -- and the reason for not implementing in v1.0 -- is the transformation functions. The simple fix would be to NOT include transformation functions (i.e. make them all linear) and apply any transformations to input prior to exporting; this breaks down when the same input is included in different slices that may have different transformation functions. While we would advise against duplicating inputs across the model (particularly using different transformation functions), and throw a warning to the user when duplicating inputs, this particular situation is present within the format_C.csv example file. The "simple" fix, therefore would break unless we detected this situation and created new columns with alternately transformed inputs. Off the top of my head, I cannot see an easy way to map the used transformation functions (if possible) to the GUI options. And what would we do if users supplied function(s) not available in the GUI? Thoughts?

how to pass the "adjusted" parameters to txpweight and txpmodel

Dear Professor,
When I was using toxpiR, I tried to modify the parameters to make txpweights have a weight ratio of 2,2,1. However, at the same time, I do not want to normalize my data, but I do not know how to pass the "adjusted" parameters to txpweight and txpmodel. Besides, how to change the name of legend and present the title of each pie chart in two lines. Do you have any ideas? My code is as follows, urgently seeking your help. Thank you a lot!

library(toxpiR)
library(ghibli)
setwd("E:\R\toxpi")
your_data <- read.csv("1.csv")
colnames(your_data) <- c("name", "I", "P", "F")
f.slices <- TxpSliceList(Slice1 = TxpSlice("I"),
Slice2 = TxpSlice("P"),
Slice3 = TxpSlice("F"))

f.model <- TxpModel(txpSlices = f.slices,
txpWeights = c(2,2,1)
)

f.results <- txpCalculateScores(model = f.model,
input = your_data,
id.var = 'name' )
txpWeights(f.results,adjusted = FALSE) <- c(2,2,1)
txpSliceScores(f.results)
library(grid)
top_50_results <- head(f.results, 50)
p1<-plot(top_50_results)

Add support for missing-ness

Add functionality to track missing-ness in TxpResult object.

• Add txpMissing slot to TxpResult class; observation by slice matrix (same as txpSliceScores ) with the proportion of missing observations
• Update txpCalculateScores to calculate missing-ness
• Add accessors for txpMissing
• Update man page and create examples showing accessors
• Add tests to cover new missing-ness functionality

Require R>=4.0

Based on some feedback from reviewers, there are nondescript issues with installing the package with lower versions of R.