Spatial Co-expression Modules Identification

To identify spatially organized gene expression patterns, we employed the NNMF package to perform non-negative factorization on the spatial gene expression matrix, then grouped gene into modules using hierarchical clustering.

Rscript run_nnmf.R --path ${prefix} --noSignatures 15 --k 6 --ntop 50

• Input:

  • cnts_train_seed_1.csv: count matrix used for NNMF.

  • locs.csv: spot location matrix paired with cnts_train_seed_1.csv.

  • gene-names.txt: name of genes selected for clustering and enhancement.

• Parameters:

  • ${prefix}: directory to the folder containing the files, i.e. data/.

  • --noSignatures: number of signatures selected after factorization.

  • --k: number of modules obtained after hierarchical clustering.

• Output: gene-names-group.txt: spatial co-expression modules identified.