Table 1. Open in new tab Computation...

	Dataset (dimension)
nipalsMCIA	2.3	15.32	289.46	671.78
MOGSA	0.53	519.55	NA	NA
MOGSA (fast svd)	0.38	434.84	13 840.66	DNC^b
Omicade	2.66	1089.53	NA	NA

	Dataset (dimension)
nipalsMCIA	2.3	15.32	289.46	671.78
MOGSA	0.53	519.55	NA	NA
MOGSA (fast svd)	0.38	434.84	13 840.66	DNC^b
Omicade	2.66	1089.53	NA	NA

Table 1.

Computation time (in seconds) comparison for different MCIA implementations and datasets.

Implementation	21 × (12 895, 547, 7016)	4193 × (2000, 32)	4193 × (33 538, 32)	1 45 810 × (2000, 35)
	Dataset (dimension)
	NCI60	10x single cell (filtered)	10x single cell (full)^a	HCA single cell^a
nipalsMCIA	2.3	15.32	289.46	671.78
MOGSA	0.53	519.55	NA	NA
MOGSA (fast svd)	0.38	434.84	13 840.66	DNC^b
Omicade	2.66	1089.53	NA	NA

Implementation	21 × (12 895, 547, 7016)	4193 × (2000, 32)	4193 × (33 538, 32)	1 45 810 × (2000, 35)
	Dataset (dimension)
	NCI60	10x single cell (filtered)	10x single cell (full)^a	HCA single cell^a
nipalsMCIA	2.3	15.32	289.46	671.78
MOGSA	0.53	519.55	NA	NA
MOGSA (fast svd)	0.38	434.84	13 840.66	DNC^b
Omicade	2.66	1089.53	NA	NA

Due to the slow performance of MOGSA without the fast svd option and Omicade, only nipalsMCIA and MOGSA with the fast svd option were tested for the larger single-cell datasets.

MOGSA (fast svd) ran into a memory error and did not converge (DNC).

This Feature Is Available To Subscribers Only