Error: More dimensions specified in dims than have been computed

[Lakshya3141]

Hello!

I am trying to implement chooseR on my scRNA dataset containing about 10k cells. I pre-processed the cells using Seuratv3, performed normalisation, scaling, PCA and umap (using default parameters).

After this, I saved an .rds file and accordingly changed the readRDS function in file1 of your package, and set npcs to 32 (my PCA has 50 dimensions to it, as evident from the elbowplot). When I try to run the subsequent for loop which iterates over my custom defined resolution vector, I get an error:

I tried various things such as not performing umap (should not affect final results thought), changing default paramters, etc but nothing works, How should I resolve this issue?

[SimoniMD]

I get this same problem, and looking for any solutions. Had tried recalculating the PCAs in the step right before, and it still bounces back the same error.

[auesro]

Did you change this part of the script?:

# Define the number of PCs to use, and which assay and reduction to use.
# We recommend testing a broad range of resolutions
# For more on picking the correct number of PCs, see:
# https://satijalab.org/seurat/v3.1/pbmc3k_tutorial.html
npcs <- 100

[nicolaromano]

I just got the same error.
Aside from what @auesro says, you should also change the following bit.
Essentially npcs was set but never used, and defaults to the one set in multiple_cluster (100)

results <- multiple_cluster(
                              obj,
                              n = 100,
                              size = 0.8,
                              npcs = npcs,  # <----- ADD THIS
                              res = res,
                              reduction = reduction,
                              assay = assay