39
OCTOBER 16-18, 2011 • PORTLAND, OR
P3
Critical Assessment of Cell Population
Identifcation Techniques for Flow Cytometry
Data: Results of FlowCAP-1
Ryan Brinkman
1
, Nima Agheepour
1
, Richard
Scheuermann
2
, Raphael Gottardo
3
, Tim Mosmann
4
1
British Columbia Cancer Agency, Vancouver, BC,
Canada,
2
University of Texas Southwestern, Dallas,
TX, USA,
3
Fred Hutchinson Cancer Research
Center, Seattle, WA, USA,
4
Univeristy of Rochester,
Rochester, NY, USA
Traditional methods for fow cytometry (FCM)
data processing have relied on subjective manual
gating to defne cell populations for statistical
analysis. Recently several groups have developed
computational methods for identifying cell populations
in multidimensional FCM data, potentially obviating
the need for manual gating. To compare the
performance of these methods, the Flow Cytometry:
Critical Assessment of Population Identifcation
Methods (FlowCAP) challenge was established.
Thirty-six analysis result submissions from fourteen
research groups were received for the frst FlowCAP
competition (FlowCAP-1). Several parametric and
non-parametric clustering methods performed well in
comparison with manual gating by domain experts
as the gold standard, using statistical measures of
algorithm performance. Combining results using
a computational ensemble method yielded further
performance improvements. These results suggest
that automated computational methods have reached
a level of maturity and accuracy such that they are
poised to replace manual gating for routine FCM data
analysis.
POSTER ABSTRACTS