Do we have scanpy version of NICHS

[HelloWorldLTY]

Hi, thanks for your great work. Considering the recent bugs in Seurat v5 as well as its complicated structure to load spatial data, do you have any plans to develop a python-based version with scanpy? Thanks.

[HelloWorldLTY]

Furthermore, I met an error when running NICHS:

Error in graph.adjacency.dense(adjmatrix, mode = mode, weighted = weighted, : long vectors not supported yet: ../../src/include/Rinlinedfuns.h:537

This happens when running:

NICHES_output <- RunNICHES(object = brain,
                           LR.database = "fantom5",
                           species = "human",
                           assay = "RNA",
                           position.x = 'x',
                           position.y = 'y',
                           k = 4, 
                           cell_types = "label",
                           min.cells.per.ident = 0,
                           min.cells.per.gene = NULL,
                           meta.data.to.map = c('orig.ident','label'),
                           CellToCell = F,CellToSystem = F,SystemToCell = F,
                           CellToCellSpatial = F,CellToNeighborhood = F,NeighborhoodToCell = T)

[msraredon]

1. NICHES can handle matrix input and also be set to output matrices, this allows cross-compatibility with python environments. Currently, NICHES requires seurat to run, though we are working on a second-generation release which will no longer have this dependency. We have not tried to implement in python, though we would be happy to collaborate to do so!
2. The new release may fix this issue, as we have implemented a new and far more efficient approach to Knn computation