Problem generating mds plot after repOverlap

[peter-yufan-zeng]

🐛 Bug

Hi There, I am running into an error while trying to graph with vis after repOverlap, exactly as detailed in the manual

To Reproduce

Steps to reproduce the behavior:

imm_ov1 = repOverlap(immdata.opc$data, .method = "public", .verbose = T)
vis(repOverlapAnalysis(imm_ov1, "mds+kmeans"))

it gives an error .by="cluster", .meta=immdata.opc$meta

However, when I run vis(repOverlapAnalysis(imm_ov1, "tsne"))

there is no problem.

This is what the beginning of print(imm_ov1) looks like

I think it might be due to the NAs in the middle. Thoughts?

Thanks in advance for your help

Expected behavior

Additional context

[vadimnazarov]

Hi @peter-yufan-zeng

sorry for the late response! Is it possible to replace NAs and try again if it works?

res = repOverlapAnalysis(imm_ov1, "mds+kmeans")
res[is.na(res)] = 0

[vadimnazarov]

Ping about the repOverlap MDS issues @peter-yufan-zeng

[peter-yufan-zeng]

Replacing the NA's still throws error

Error in MASS::isoMDS(.data, k = 2, trace = F) : zero or negative distance between objects 1 and 2

[vadimnazarov]

Thank you, will look into this as well. Thank you a lot for coming back and helping us improve the package, @peter-yufan-zeng !

[peter-yufan-zeng]

@vadimnazarov no problem. I am a huge fan of the package and appreciate your help 👍

[DCGenomics]

Having the same issue, thanks!

[Manikgarg]

Same for me.

[parvathisudha]

Hi,

I am having the same issue. Is there any solution for this error?

Thanks! Parvathi.

[vadimnazarov]

Hi, I tried to reproduce it on the dev version of immunarch and test data data(immdata), but I don't have any problems. Could you share the overlap matrix here? What version of the package are you using?