buildTree Error

[02infi]

Hello @farrellja,

I am trying to run URD on single cell data which is around 50K cells. When I am running buildTree(), I am getting the embedding error :

[1] "Calculating divergence between 47 and 56 (Pseudotime 0.439 to 0.553)" Error in embed(1:n.pseudotime.groups, dimension = min(window.size, length(unique(visit.data$pseudotime.group)))) : wrong embedding dimension

Does this mean that all the cells in the cluster have unassigned pseudotime ?

[farrellja]

@02infi

Sorry for that other user who does not even bother to read documentation bothering you.

It's possible that what you described occurred. Unfortunately, it will be a bit difficult to check, as I suspect that segments 47 & 56 are created during the process of building the tree, and if it's not completing it will be tough to identify exactly which cells the are.

Can you provide a reproducible example that is your workspace right before running buildTree and then the command you issue to buildTree?

[02infi]

@farrellja

Well the segment 56 is created joining the segment 3 and 4. Now iteratively it trying to calculate the divergence between newly created segment and the already present tip (segment 47).

URD_object.built <- buildTree(object = URD_object, pseudotime = "pseudotime",tips.use = tips.to.use,save.breakpoint.plots = F, verbose = T,cells.per.pseudotime.bin = 80)

I follow the documentation step by step, excluding the some tips according to diffusion maps. I also face the same problem when I am trying to rerun URD on zebrafish data from your lab. It also gives me the same error.

Do you think the parameter tuning can overcome this error ?

Thanks!

[farrellja]

@02infi

It's especially surprising if you're having the same issue with our data — the software obviously has run OK on that data set in the past ;) It's possible that parameter tuning would avoid the error, but it seems like something else is happening — either one of your inputs has an issue, or there is a bug in our code that only occurs under certain conditions.

First, please try re-running with the most up-to-date version of URD (1.0.3.9003: devtools::install_github("farrellja/URD@debug").

If that doesn't work, then can you share your workspace from right before running the above command? You can use the R command save and then share the file with me via Dropbox, Google Drive, or some other such method. It will be hard for me to help you troubleshoot otherwise.

[02infi]

@farrellja

Thanks for the debug file. I will rerun it and get back to you. Yes, I will share the URD object too.

Regarding the zebrafish data from your lab, do you think If I rerun it with same exact parameters, I will expect the same global structure of development ?

Finally, thanks for building this package. I really enjoying it :)

[xiatianjihao]

@02infi , Could you have solve the problem? I have met the same issue with you.