Week # 8 - General Discussion Thread

[UMGCCCFCSR]

This is the Week # 8 - Manual and Automated Gating discussion thread.

I fully expect most of you to deep dive the various gating approaches, and encounter all sorts of unique and interesting edge cases where my explanation breaks down. So this is the place to discuss them if they don't rise to the level of their own discussion post.

Among the anticipated ones, how to use the +/- and * system for openCyto, changing dimensions on flowGate, and better ways to plot visuals than autoplot.

Regardless, have at it and happy Gating day!

Best-
David

[minhai2]

Hi David,

Just a quick question about "install_github("DavidRach/flowGate")", does this only work for R >= 4.5? I'm currently using R 4.4.3, and it seems like I cannot install that package : (

Thanks,
Min

[DavidRach]

Hi @minhai2, yes, but since it's my forked version I went ahead and rolled back the requirement version to 4.4, so go ahead and try to install_github again.

Not to jump ahead to Week 23, the error you are getting is due to the requirement that is listed in an R packages description file. The forked flowGate version looks like this, relavent line under depends

Both R and Bioconductor undergo twice yearly releases (one major, one minor), so often times packages set the requirement to match the existing release (even though they will often work for the version before without issue).

The main issue you might still encounter is if one of the flowGate dependency packages also requires 4.5, you may get the same error for it. But let's kick the can down the road if/until that happens😄

[minhai2]

Thank you, David!

[carloscgomiz]

Hi!

I have a question regarding OpenCyto. I imagine we will get more into depth in the following weeks, but I am having issues when applying this gating to different samples. I imported two different fcs files, and no matter how I modify the parameters in the template, if I get them to fit one sample, the gate gets completely different in the other one:


See the FSC-A vs SSC-A and the LiveDead Blue plots.

Is there any way to coerce the gates to be the same in all samples?

Thank you in advance!
Carlos.

[lkisselbach-afk]

Hi Carolos & David,

I'm having a similar issue when trying to apply the same gate over multiple samples in OpenCyto.

Did you find a solution to this yet?

Thanks,
Lynn

[DavidRach]

Hi @carloscgomiz & @lkisselbach-afk, sorry for the delay!

Part of the issue I realized is the autoplot numbers that appear on the x/y axis are not the actual values needed when using gate_ranges, at least in the context of fluorophores. See this week's subsection for several visualization options where the numbers you see match those openCyto uses for setting gate ranges.

As far as coercing them to be more fixed across all samples, that would be for min_density providing a very narrow gate range (think "gate_range=c(3001, 3002)").

As far as the lymphocyte gate misbehaving, its using flowClust.2d() function in the background, so we would need to find the arguments for that. I would start here before working over to the function code and parsing arguments since there is no help file for it specifically.

Best-
David

[Rhodnius]

Hi David (and colleagues),

Thanks again for all your hard work organizing the course!

Changing subjects, because I am a bit late with everything, I am diving into week 8 lesson and I found an error that most probably someone already had.

When I try installing flowGate I have an error trying both with "remotes" or "pak"

remotes

install_github("DavidRach/flowGate")

Using GitHub PAT from the git credential store.
Error: Failed to install 'unknown package' from GitHub:
HTTP error 401.
Bad credentials

pak

pkg_install("DavidRach/flowGate")

Error:
! error in pak subprocess
Caused by error:
! Could not solve package dependencies:

• DavidRach/flowGate: ! pkgdepends resolution error for DavidRach/flowGate.
  Caused by error:
  ! Bad GitHub credentials, make sure that your GitHub token is valid.
  Caused by error in stop(http_error(resp)):
  ! Unauthorized (HTTP 401).
  Type .Last.error to see the more details.

Any suggestions to get the package installed? Did the flowGate mantainers replied to your PR?

Thanks!

[DavidRach]

Hi @Rhodnius, can you try to install another package via either remotes or pak (try "saeyslab/CytoBright").

If you are not able to install, it is likely that your GitHub credential (probably a token if you are still using Rstudio) expired and needs to be refreshed (here is a similar error message).

Best-
David

PS: Still nothing on the flowGate maintainer side.

[Rhodnius]

Hi David, thanks for your suggestion. I solved the problem by refreshing the git token in Rstudio.

What I did:

1. Generated a new token: usethis::create_github_token(), copied the generated string (without closing the tab to guarantee saving the token

2. Store it: I ran gitcreds::gitcreds_set(). When prompted, paste the token you just copied.

3. Restarted R: This is crucial to ensure the new credentials are loaded.

4. Cleaned .Renviron Run usethis::edit_r_environ().

Have a nice day,

Juan C.

[DavidRach]

Glad it worked @Rhodnius, and thanks for the solution. Will add it to the common issues documentation page I have been working on (compiling various issue threads). Best- David

[MacarenaLM]

Hi David,
I am starting with this week and I have a doubt. In your screenshots, the version of the flowgate is 0.99.2. The version that I have installed is 1.10.0. It is still recommended to remove the package and reinstall again the forked version? Many thanks in advance!
Regards,
M.

[DavidRach]

Hi @MacarenaLM, you likely got the Bioconductor version still installed on your system. This was the one that the shiny app was bugged out due to the ggplot2/ggcyto changes. Still haven't heard from the developers, which was why at the beginning of Week 08 walkthrough I have everyone remove and install via remotes the course forked version where the fixes were made.

remove.packages("flowGate")

# Close and reopen positron, then 

remotes::install_github("DavidRach/flowGate")

Hope this helps!
David