readVizgen, multiple z-planes

[heziqing]

Hi, I am using readVizgen to read merfish data. The input files are as follows:

The cell_boundaries.parquet file has multiple z-planes.

When I used the readVizgen function to read the MERFISH data, an error occurred.
sfe_mer <- readVizgen(dir_use, z = "all", image = "DAPI", add_molecules = TRUE,flip = "geometry", BPPARAM=BiocParallel::MulticoreParam(14, tasks = 80L,force.GC = FALSE, progressbar = TRUE))

sfe_mer <- readVizgen(dir_use, z = 3L, image = "DAPI", add_molecules = TRUE,flip = "geometry", BPPARAM=BiocParallel::MulticoreParam(14, tasks = 80L,force.GC = FALSE, progressbar = TRUE))
The error is:
Sanity checks on cell segmentation polygons:

..found 33680 cells with (nested) polygon lists
..applying filtering
..removing 25 empty polygons
Casting MULTIPOLYGON geometry to POLYGON
|============ | 18%
Error: BiocParallel errors
14 remote errors, element index: 4, 284, 571, 867, 1140, 1421, ...
22269 unevaluated and other errors
first remote error:
Error in st_geometry(polys[polys[[cell_ID]] %in% dupl_cells[i], ])[[which_keep]]: subscript out of bounds
In addition: Warning message:
In .filter_polygons(polys, min_area = 15, BPPARAM = BiocParallel::MulticoreParam(14, :
There are 53183 cells with multiple pieces in cell segmentation larger than min_area, whose first 10 indices are: 3, 4, 5, 9, 11, 13, 14, 15, 16, 17. The largest piece is kept.

How can I resolve this issue? Thanks in advances!

[heziqing]

Even I filtered the cell boundaries by the third plane, the same error still occured.

[alikhuseynov]

Hi, could you please install this package using devel branch und see if the same error occurs?
Thanks

[heziqing]

Thank you for your response. I will try the version you mentioned. However, I’d like to ask which function can be used to visualize molecules across all planes, similar to Seurat's function:
ImageDimPlot(vizgen.obj, molecules = "Slc17a7", nmols = 10000, alpha = 0.3, mols.cols = "red").
It seems that in Seurat, ImageDimPlot can only display molecules from a specific plane, such as those on the third layer.

[alikhuseynov]

Thank you for your response. I will try the version you mentioned. However, I’d like to ask which function can be used to visualize molecules across all planes, similar to Seurat's function:

ImageDimPlot(vizgen.obj, molecules = "Slc17a7", nmols = 10000, alpha = 0.3, mols.cols = "red").

It seems that in Seurat, ImageDimPlot can only display molecules from a specific plane, such as those on the third layer.

Yes, that is in Voyager, take a look at these recent workflows:
Xenium ->
https://pachterlab.github.io/voyager/articles/vig5_xenium.html
Vizgen ->
https://pachterlab.github.io/voyager/articles/vig6_merfish.html

[lambdamoses]

I can reproduce this error on an example dataset from Vizgen. I think I found the culprit. It has nothing to do with z-planes. It's from matching integer64 cell IDs. The subscript is out of bound when when we get a wrong match using integer64; base R can only handle integer32. I edited the code to match using a string ID and it eliminated that error. See if it works for you.