o Uݢg@sdZddlmZddlmZddlZedd d Zeddd Zddd Ze dd ddZe dd!ddZ dd"ddZdS)#zJFunctions for fast, memory-efficient operations on masked sparse matrices.) annotations)overloadNrow_mask#np.ndarray[int, np.dtype[np.bool_]]col_maskaxisNonereturn np.genericcCdSNmatrixrrrr r `/oak/stanford/groups/akundaje/marinovg/programs/cellranger-9.0.1/lib/python/cellranger/sparse.py sum_maskedrint%np.ndarray[int, np.dtype[np.generic]]cCr r r rr r rrr int | NonecCs|dksJt||jdksJt||jdksJ|jdks%J|jdks,J|dkr;|jdd||S|dksC|dur[|||}|dkrP|S|dusVJt|Std) a"Sum a masked sparse matrix along an axis. Args: matrix (scipy.sparse.csc_matrix): Matrix axis (int): Axis to sum away; None for total sum. row_mask (np.array): Boolean mask for rows col_mask (np.array): Boolean mask for columns Returns: np.array cscrboolFcopyNzaxis must be 0, 1, or None) getformatlenshapedtype transposedotnpsum ValueError)rrrrrow_sumsr r rr"s  @ threshold chunk_size_mb np.uint64cCr r r mrrr&rr'r r rcount_ge_maskedHr+$np.ndarray[int, np.dtype[np.uint64]]cCr r r r)r r rr+Sr,cCs|dksJ|dksJt||jdksJt||jdks$J|jdks+J|jdks2J|jjj}|jjj}||}t|j } | |} |d} dg} d} t t| D]}| | |7} | | krl| |d} qYg}|dvrwt dt | D]D\}}|t| dkr| d|n|jd}t||}|dd|f|ktj}|dkr|jd d |}n|||}| |q{|dkrt||S|dkrt||S|durtt||St d) aCount values greater than or equal to a threshold in a masked sparse matrix. Args: m (scipy.sparse.csc_matrix): Matrix axis (int): Axis to sum away; None for total count. threshold (int): Only count elements exceeding this threshold row_mask (np.array): Boolean mask for rows col_mask (np.array): Boolean mask for columns chunk_size_mb (int): Operate on chunks of this size; don't copy the entire matrix at once. Returns: np.array rrrri)rrNzAxis must be 0, 1, or NoneNFr)rrrrindicesitemsizedatar!diffindptrrangeappendr# enumeratesliceastypeuint64rr concatenater")r*rrr&rr'Z ind_bytesZ data_bytesZbytes_per_nz_elemZnz_elem_per_colZ bytes_per_colZchunk_size_bytesZ chunk_starts chunk_sizeiresultsZ chunk_idxZ chunk_start chunk_endchunkZ submatrixsr r rr+^sN      $   )rrrrrrr r )rrrrrrr r)rrrrrr)r%) rrrrr&rrrr'rr r() rrrrr&rrrr'rr r-) rrrrr&rrrr'r) __doc__ __future__rtypingrnumpyr!rr+r r r rs"     &