Several temporal features including the time of mitotic arrest, cell death or cell cycle arrest, or the duration of mitosis were not quantified. In principle, the nature of the data time lapse movies of dividing cells asks for analysis of the single cell tracking graphs. However, reliable tracking thenthereby of the cells used in this experiment requires a time resolution between image frames lower than 10 min. For the main Mitocheck data set, the decision had been made to use a lower tem poral sampling frequency ?1 in order to allow for a larger volumes in other dimensions of the experi mental design, in particular, number of siRNAs tested and number of cells per siRNA. In other experiments, there may be analogous considerations that hinder tracking at the single cell level, while still providing population level time course data.

In this study, we used a cell population level dynamic model to represent the temporal evolution of dividing cells. By fitting cell counts in four transient cellular states, our model yielded parameters that quantify the dynamic effects of siRNA treatments on cell population levels. Model parameters allowed reliable estimation of the pen etrance and time of four disruption events of the cell cycle, quiescence, Anacetrapib mitosis arrest, polynucleation and cell death. We also derived the interphase and mitosis durations from penetrance parameters. We found 2190 siRNAs that resulted in quiescence, mitosis arrest, polynucleation or cell death at specific times, or increased interphase or mitosis duration.

Comparison of the results with known cell cycle and cell death regulators and systematic gene enrichment analysis indicate high sensitivity and accuracy of the method. The reported list is a useful resource, con taining testable hypotheses about causal roles of genes in cell cycle regulation and cell death. Results and discussion Modelling cell population dynamics We considered the cell count data from the Mitocheck pri mary screen, consisting of 206,592 movies of siRNA spot experiments targeting 17,293 genes. Most of the genes were targeted by at least two independent siRNA sequences, each done in at least three spots. Four controls were repeatedly used on each slide, siScrambled, a non targeting negative control, siKIF11, targeting the gene KIF11, which encodes a kinesin needed for centro some segregation, siCOPB1, targeting an essential protein binding to the Golgi vesicle and siINCENP, targeting a centromere associated protein coding gene required for proper chromosome segregation and cytokinesis.

Each spot experiment yielded time courses of cell counts of 16 morphologically distinct transient nuclear morpholo gies, first acquired 18 h after cell seeding, Vandetanib cancer and then measured every 30 min for 48 h. In total, more than 2 bil lion nuclear morphologies were measured and classified.