R/cgraph_gcor_norm.r
mcell_cgraph_norm_gcor.RdThis is generating a randomized umi matrix by substituting each umi call (per cell and gene) with a umi call from the same gene on a cell that is connected to the original cell in a given similarity graph. If the similarity graph is based on some specific features that express a specific process (e.g. cell cycle), this can help in normalizing this effect and retaining correlation that are independent of it.
mcell_cgraph_norm_gcor(cgraph_id, mat_id, K = -1, min_gtot = 1000)id of the cell graph to be used for normalization
the matrix with umis to be analyzed
the maxium number of edges to consider per cell in normalization. This can be smaller than the K used in cgraph_id, to provide tighter control
minimal number of umis to consider when computing gene-gene correlaitons
a list with two entries. gcor - is the gene correlation matrix of the downsampled matrix. r_gcor - is the correlation of the randomized umi matrix