sparx is a memory-efficient tool for working with sparse matrices in the Matrix Market format on-disk.
Subset a matrix on-disk:
sparx subset -i <matrix.mtx> --rows <row_index.txt> --cols <col_index.txt> -o <matrix_subset.mtx>
Compute matrix statistics (nonzero count, sum, mean, variance, standard deviation, min, max) for each row and column:
sparx stats -i <matrix.mtx> -o <filename>
To compile, clone the git repo and run cargo install:
git clone [email protected]:stuart-lab/sparx.git
cd sparx; cargo install --path .
Precompiled binaries are also available in the release