Citrus transformation warning

[selly84]

Hi, I am having warnings when running citrus on my flow cytometry data: "transformation introduced infinite values in continuous y-axis". Does anyone know how to solve this issue?
In addition, when setting up citrus it does not allow me to change the cross validation folds (by default is 1 and is not good), do you know why and how to solve this?
Thanks a lot

[SamGG]

Hi,

• 1st question: no idea. I think you should share your FCS in order to solve this question.
• 2nd question: you could change the value in the generated runCitrus.R file, and then source this script.

Alternatively, I integrated a few corrections made by other followers of Citrus, added a white background theme, added PNG plots that allow better overviews than PDF, and corrected the cross validation value problem. If you want to try, just use my repo instead of the original. Once tested thoroughly, the changes will be sent as a pull request to the main repository.

devtools::install_github("SamGG/citrus")

Best.

[rbruggner]

Very cool Sam- Looking forward to seeing the deets!

[selly84]

Hi, thanks a lot!! I will try the SamGG option to see if I get a better output.
Best

[selly84]

Dear SamGG, I tried your version and indeed some issues are solved. However I encounter another error:
package ‘Rclusterpp’ is not available (for R version 3.5.1) . Do you know how can I solve this?
Thanks a lot again!

[SamGG]

Hi. Package ‘Rclusterpp’ was removed from the CRAN repository. So you cannot install it using install.package(). You have to follow the installation procedure described at https://github.com/nolanlab/citrus/wiki/Installing-Citrus which states to use install_github for Rclusterpp. This procedure works. The main point I noticed very recently is that citrus does not manage compensation at all (@rbruggner Robert, correct me if I am wrong). Because it was designed for mass cytometry. That's may be the explanation for your 1st error.
Best.

[SamGG]

By the way, if you tried "my" version that's means you already got Rclusterpp installed and running. So the error your report about Rclusterpp on R3.5.1 relates to an update or a new installation. Correct?