spatial() assigns multiple samples to the same plate position in the same batch

[AnnaPagotto]

Hi Lluís, thank you for this amazing package. I'm using it for my project, but I encountered some issues when distributing the samples in the plate grid. I have some cases where spatial() assigns the same position in the same batch to two samples.
Could you help me with that?

Thank you a lot!
Best,
Anna

[llrs]

Hi Anna, happy to hear from a happy user.

Sorry to hear that spatial doesn't fully work. I think I tested to prevent this from happening, but I will give it a look. Could you provide the code where this happens? This way I will make sure it won't happen again.

[AnnaPagotto]

Hi Lluis, Thank you very much for the super fast reply. Here is the code and the dataset. Thanks! Anna Il giorno mar 5 dic 2023 alle ore 12:15 Lluís Revilla < ***@***.***> ha scritto:

…

Hi Anna, happy to hear from a happy user. Sorry to hear that spatial doesn't fully work. I think I tested to prevent this from happening, but I will give it a look. Could you provide the code where this happens? This way I will make sure it won't happen again. — Reply to this email directly, view it on GitHub <#45 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AYJGFB5GHRC32OUCVBNILB3YH363RAVCNFSM6AAAAABAHO4A5WVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQNBQGU2DEOJXGQ> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[llrs]

Sorry, if you send the data via email, I don't see it in the github issue. But don't worry that I found a way to reproduce the issue:

data(survey, package = "MASS")
set.seed(4568)
nas <- c(137, 70)
index <- replicates(survey[-nas, c("Sex", "Smoke", "Age")], size_subset = 50,
                   iterations = 25, controls = 15)
index2 <- spatial(index, survey[-nas, c("Sex", "Smoke", "Age")], iterations = 25)
uniquness <- vapply(index2, function(x){length(unique(x)) == length(x)},
                    FUN.VALUE = logical(1L))
expect_false(any(!uniquness))

i1 <- inspect(index, survey[-nas, c("Sex", "Smoke", "Age")])
i2 <- inspect(index2, i1, index_name = "spatial")
# This should be empty too
i2 |> group_by(batch) |> count(spatial) |> ungroup() |> arrange(batch) |> filter(n!= 1)

[AnnaPagotto]

Oh, sorry, I didn't realise it.
Let me know if I can provide with anything else. I can still send you the code if you wish.

[llrs]

Hi Anna, I fixed this locally, thanks for opening the issue. However, I found that there is a problem with batch_names which affects inspect too when there are controls or repeated positions. Do you have controls in your data? Or would this be enough for the moment?

[AnnaPagotto]

Hi Lluis, I'm assigning the position to my controls separately, hence this should be enough for now. Thank you a lot for your work!

…

On Thu, Dec 7, 2023, 11:24 Lluís Revilla ***@***.***> wrote: Hi Anna, I fixed this locally, thanks for opening the issue. However, I found that there is a problem with batch_names which affects inspect too when there are controls or repeated positions. Do you have controls in your data? Or would this be enough for the moment? — Reply to this email directly, view it on GitHub <#45 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AYJGFB6ZUHO7FPZHSAMHWQLYIGKMZAVCNFSM6AAAAABAHO4A5WVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQNBVGA3TKOJTHE> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[AnnaPagotto]

Hi Lluis,
I removed and reinstalled the package. I thought that was sufficient to have the fixed version, but the bug is still there. What do you mean by "fixed locally"? Sorry for my ignorance.
Anna

[llrs]

I meant that it was only in my computer, not yet on github or CRAN.

To install the package with the fix you will need to use remotes::install_github("llrs/experDesign") and it will install the development version with the fix. You don't need to uninstall it and reinstall it again, but you will need to close R and start again R after the installation to be sure that the new version is used.

If it is not fixed then I would need to know the code used and the experDesign version used.

[AnnaPagotto]

Thanks, the positions are now uniquely assigned. I just noticed now something that was not there before the fixing: I have two numeric and two categorical variables. For one categorical variable (in my case "age") it doesn't distribute the samples randomly in the plate.

Here the example:

[llrs]

Nice plot! I assume you blocked/omitted the first column? Or you only provided with columns 2 to 8?

I think I reverted that change, it slipped through. You will need to repeat the steps in #45 (comment).
I am not sure how you have age as categorical, in the plots it is displayed as numeric. Did you use cut to categorize it or it is a simple typo?

[AnnaPagotto]

Sorry, it was a typo, I have age as numeric indeed.
PS: yes I blocked the first and the last two columns. And I used ggplate for the plot :)

[AnnaPagotto]

It is perfectly working now! Thank you a lot!

[llrs]

Glad it works now! I hope it helps well designing your experiment!
Thanks for the tip about ggplate! I might use it in the future.