flowEMMi: an automated model-based clustering tool for microbial cytometric data

BMC Bioinformatics

Table 3 Comparison of automated clustering approaches

Tool	Running time (h:mm:ss) ^a	Output features
		Determine number of clusters	Shape of clusters	Separate background	Calculate cell numbers
flowEMMi	0:05:31	yes	ellipsoid	yes	yes
flowFP	0:00:03	no	rectangular	no	not applicable
SamSPECTRAL	0:06:25	space-part.	arbitrary	no	not applicable
flowDensity	0:00:02	no	rectangular	no	not applicable
flowMeans	0:00:17	space-part.	non-spherical	no	not applicable
flowClust	1:15:30	yes	ellipsoid	no	yes
flowMerge	(Table 4)	yes	ellipsoid	yes	yes
FLAME	−^∗	−^∗	−^∗	−^∗	−^∗

Automated approaches were compared regarding the running time and the abilities to identify rare cell types, to separate cell clusters from background clusters and to calculate the real cell numbers for each cell cluster. ^a Running time calculated on a Intel(R) Core(TM) i5-3210M CPU @ 2.5 GHz with 4096MB RAM and Windows 7 Enterprise 64-Bit Edition. FLAME: “ −^∗” denotes that no results were received as our submitted “jobs” were always in the queue for several days and later cancelled by the server. flowEMMi is the implementation discussed in this work. space-part. denotes k-means type algorithms that do not produce tight clusters

ISSN: 1471-2105