Cellular heterogeneity generated by multipotent cell growth is characterised by single cell analysis. Continuous tracking of single cells and their progeny provides prospective data that can establish a causative link between molecular changes and competing cell fate outcomes. Examples of (mutually exclusive) competing cell fates are mitosis/apoptosis/quiescence or stem cell self-renewal/differentiation. The competing fate problem is stated as follows: How does one quantify the risk of a cell fate outcome if its observation is censored by a competing fate? Recently statisticians were motivated to solve the competing risks problem for the clinical context, for example to select optimal patient cancer therapies when there are mutually exclusive treatment outcomes [1, 2]. These powerful statistical methods are translated to the field of cell biology for the first time.
Regression and cluster analysis for competing risks is performed using the R packages developed by Fine and Scheike. In-house semi-automated cell tracking software was implemented in Matlab. The application of competing risks statistics to analyse prospective single cell data is illustrated using collaborators experimental data: a) hESC cell cycle analysis using the FUCCI reporter (Draper et al, McMasters Stem Cell and Cancer Research Institute, Hamilton, Canada), b) the effect of G-CSF and M-CSF on granulocyte-macrophage progenitor development (Schroeder et al. ETH, Zurich) and c) analysis of microenvironment on the fate of cardiac-derived, MSC-like cells (Harvey et al. Victor Chang Cardiac Research Centre, Sydney). We conclude that competing risks statistical inference on prospective single cell data is an important scientific method for stem cell research.
1. Scheike, T.H. and M.J. Zhang, Flexible competing risks regression modeling and goodness-of-fit. Lifetime Data Analysis, 2008. 14(4): p. 464-483.
2. Zhou, B.Q., J. Fine, A. Latouche, and M. Labopin, Competing risks regression for clustered data. Biostatistics, 2012. 13(3): p. 371-383.