Pattern recognition in high-content cytomics screens for target discovery - Case studies in endocytosis

Finding patterns in time series of images requires dedicated approaches for the analysis, in the setup of the experiment, the image analysis as well as in the pattern recognition. The large volume of images that are used in the analysis necessitates an automated setup. In this paper, we illustrate the design and implementation of such a system for automated analysis from which phenotype measurements can be extracted for each object in the analysis. Using these measurements, objects are characterized into phenotypic groups through classification while each phenotypic group is analyzed individually. The strategy that is developed for the analysis of time series is illustrated by a case study on EGFR endocytosis. Endocytosis is regarded as a mechanism of attenuating epidermal growth factor receptor (EGFR) signaling and of receptor degradation. Increasingly, evidence becomes available showing that cancer progression is associated with a defect in EGFR endocytosis. Functional genomics technologies combine high-throughput RNA interference with automated fluorescence microscopy imaging and multi-parametric image analysis, thereby enabling detailed insight into complex biological processes, like EGFR endocytosis. The experiments produce over half a million images and analysis is performed by automated procedures. The experimental results show that our analysis setup for high-throughput screens provides scalability and robustness in the temporal analysis of an EGFR endocytosis model. © 2011 Springer-Verlag.