filterProbability() fails or hangs with large SpatialCellChat

[Chiaraderk]

Hey, I'm currently running new script for SpatialCellChat (CellChat V3) on a large 10x Spatial Trascriptimics HD 3’ dataset and encountering issues with filterProbability(). My chat.rds object is ~24.9 GB, and the function either hangs indefinitely or crashes, even after increasing memory limits.

Code used:
options(future.globals.maxSize = 220 * 1024^3)
plan(multisession, workers = 8)
chat <- filterProbability(chat)
#> >>> Filter out non-significant communication with a probability quantile being 0.95 for each L-R pair...
#> ℹ Future strategy in use: parallel with 2 workers
#> ✔ Filtering is done.
chat <- filterCommunication(
chat, min.cells = NULL,
min.links = 10,
min.cells.sr = 10
)

Is there a recommended strategy for large datasets?
Thanks!

[Ethel-0]

For large datasets, steps involving significance testing tend to consume substantially more computational resources. Optimizing the computational cost of Spatial CellChat to better accommodate larger datasets is part of our ongoing plans. At present, you can use the makeGridSpatialCellChat() function to appropriately reduce the dataset size and make computations more manageable; more details about this function can be accessed via help().

In addition, we have updated the code to ensure that the progress bar is displayed. If you find that some time-consuming steps do not display a progress bar, please try to reinstall the latest version from GitHub.