A framework for evaluating the performance of SMLM cluster analysis algorithms
Nat Methods. 2023-02-01; 20(2): 259-267
DOI: 10.1038/s41592-022-01750-6
Lire sur PubMed
Nieves DJ(1)(2), Pike JA(2)(3), Levet F(4)(5), Williamson DJ(6), Baragilly M(1)(7), Oloketuyi S(8), de Marco A(8), Griffié J(9), Sage D(10), Cohen EAK(11), Sibarita JB(4), Heilemann M(12), Owen DM(13)(14)(15).
Author information:
(1)Institute of Immunology and Immunotherapy, College of Medical and Dental
Sciences, University of Birmingham, Birmingham, UK.
(2)Centre of Membrane Proteins and Receptors (COMPARE), University of
Birmingham, Birmingham, UK.
(3)Institute of Cardiovascular Sciences, College of Medical and Dental Sciences,
University of Birmingham, Birmingham, UK.
(4)Interdisciplinary Institute for Neuroscience, CNRS, IINS, UMR 5297,
Université de Bordeaux, Bordeaux, France.
(5)Bordeaux Imaging Center, CNRS, INSERM, BIC, UMS 3420, US 4, Université de
Bordeaux, Bordeaux, France.
(6)Department of Infectious Diseases, School of Immunology and Microbial
Sciences, King’s College London, London, UK.
(7)Department of Mathematics, Insurance and Applied Statistics, Helwan
University, Helwan, Egypt.
(8)Laboratory of Environmental and Life Sciences, University of Nova Gorica,
Rožna Dolina, Slovenia.
(9)Laboratory of Experimental Biophysics, Institute of Physics, Ecole
Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland.
(10)Biomedical Imaging Group, Ecole Polytechnique Fédérale de Lausanne (EPFL),
Lausanne, Switzerland.
(11)Department of Mathematics, Imperial College London, London, UK.
(12)Institute of Physical and Theoretical Chemistry, Goethe-University
Frankfurt, Frankfurt, Germany.
(13)Institute of Immunology and Immunotherapy, College of Medical and Dental
Sciences, University of Birmingham, Birmingham, UK. .
(14)Centre of Membrane Proteins and Receptors (COMPARE), University of
Birmingham, Birmingham, UK. .
(15)School of Mathematics, University of Birmingham, Birmingham, UK.
.
Single-molecule localization microscopy (SMLM) generates data in the form of
coordinates of localized fluorophores. Cluster analysis is an attractive route
for extracting biologically meaningful information from such data and has been
widely applied. Despite a range of cluster analysis algorithms, there exists no
consensus framework for the evaluation of their performance. Here, we use a
systematic approach based on two metrics to score the success of clustering
algorithms in simulated conditions mimicking experimental data. We demonstrate
the framework using seven diverse analysis algorithms: DBSCAN, ToMATo, KDE,
FOCAL, CAML, ClusterViSu and SR-Tesseler. Given that the best performer depended
on the underlying distribution of localizations, we demonstrate an analysis
pipeline based on statistical similarity measures that enables the selection of
the most appropriate algorithm, and the optimized analysis parameters for real
SMLM data. We propose that these standard simulated conditions, metrics and
analysis pipeline become the basis for future analysis algorithm development and
evaluation.
© 2023. The Author(s), under exclusive licence to Springer Nature America, Inc.