Parallelized Matrix Multiplication for kNN Edge Weighting
Usage
cell_knn_matrix_multiplication_parallel(
knn,
cell_feature_mat,
feature_feature_mat,
chunk_size = 5000
)
Arguments
- knn
A matrix or data frame of kNN edges (columns: `i`, `j`).
- cell_feature_mat
A numeric matrix of cell-by-feature weights.
- feature_feature_mat
A sparse feature similarity matrix (e.g., clone-by-clone).
- chunk_size
Integer. Number of edges to process per parallel chunk (default: 5000).
Value
A numeric vector of computed scores, one per kNN edge.
Details
Computes edge-level scores via parallelized matrix multiplication between cell and feature matrices,
allowing scalable construction of clone-informed kNN graphs.