Update scPred for large datasets, Error: 'problem too large'

[Elhl93]

Hi,

Thanks for the great tool. I am working with a single-cell experiment of ~1 million cells, however, I can not properly integrate the Seurat object in the current workflow without downsampling, it seems that the sparse matrix is densified which causes an error:

'problem too large' in file ../Core/cholmod_dense.c, line 102

However, downsampling solves the problem. Problematic functions are for example scPredict. But as more studies are now growing in size, it might be worth considering changing the source code.