Anslational speed distributions are shown as cumulative distribution plots. Similar plots, (C) and (D), depict turn speed data. (E) Cumulative distribution plot of track meandering index distributions. (F) Mean squared displacements for offered durations (anyplace inside the temporal domain, not from time zero only) plotted on LY2365109 (hydrochloride) log-log axes. The gradients of linear regression fitted models are offered. PubMed ID:http://www.ncbi.nlm.nih.gov/pubmed/20188782 (G) X and Y coordinates relative to starting positions of 40 tracks, selected to capture the entire range of net displacements. (H) Scatter plot of track meandering indices against duration, Spearman’s rank correlation coefficient is provided. The model’s parameter values are provided. We note that model calibration was performed using metrics of panels A, C and E only. (PNG) S24 Fig. Alignments of each simulated motility models’ Pareto front solutions’ median track translational and turn speed distributions with in vivo information. Each motility model was independently simulated and calibrated against both T cell and neutrophil data 3 times. Pareto fronts were compiled for each model from the solutions in all three calibrations. The alignment of each Pareto front solution median track translation/turn distribution with the corresponding in vivo data was assessed using the Kolmogorov-Smirnov (KS) statistic. Shown here are the distributions of KS values across all Pareto front solutions. These information broadly correspond withPLOS Computational Biology | DOI:10.1371/journl.pcbi.1005082 September 2,29 /Leukocyte Motility Assessed through Simulation and Multi-objective Optimization-Based Model Selectionthe independent statistical modeling of translation and turn speed dynamics, Fig 1. Heterogeneous CRW models better capture T cell and neutrophil dynamics than homogeneous CRW models, with the exception of neutrophil median track turn speeds, where no discernible difference is found. (PNG) S25 Fig. HeteroCRW and HomoCRW Pareto fronts plotted against each objective KS value. Calibration was against T cell data (top), or neutrophil information (bottom). `Trans’, translation speed KS values; `Turn’, turn speed KS values; `MI’, meandering index KS values. The large arrow identifies the origin, which represent a perfect reproduction of in vivo motility dynamics. Faded color dots lie further from the viewer. Plots have been rotated to emphasize the separation between the two Pareto fronts. (PNG) S26 Fig. IHeteroCRW and IHomoCRW Pareto fronts plotted against each objective KS value. Calibration was against T cell information (top), or neutrophil data (bottom). `Trans’, translation speed KS values; `Turn’, turn speed KS values; `MI’, meandering index KS values. The large arrow identifies the origin, which represent a perfect reproduction of in vivo motility dynamics. Plots have been rotated to emphasize the separation between the two Pareto fronts. (PNG) S27 Fig. Correlations between translation and turn speeds for various CRW models. IHomoCRW and IHeteroCRW explicitly prescribe a negative correlation between translation and turn speeds, motivated by finding such a correlation in our in vivo information. Information shown represent the best solutions, as determined by lowest value, for each model when calibrated against each dataset. (PNG) S28 Fig. Homo and IHomoCRW Pareto fronts plotted against each objective KS value. Calibration was against T cell data (top), or neutrophil information (bottom). `Trans’, translation speed KS values; `Turn’, turn speed KS values; `MI’, meandering index KS values. The large ar.