Bioinformatics: An Image Informatics Pipeline for High Content Screening

摘要

High content screening (HCS) has recently emerged as a promising solution to improve the quality of decision-making and lead selection in drug discovery. However, bioinformatics tools required for analyzing high throughput image data are rather immature. In this report, we review major image analysis techniques of HCS and illustrate them with examples from screening applications at our institution. The power of HCS draws from the sensitivity and resolution of automated microscopy with multiwell plates, combined with the availability of fluorescent probes that are attached to specific subcellular components, such as chromosomes and microtubules, for visualization of cell division or mitosis using epi-fluorescence microscopy techniques. The cell-based assays are hosted in multi-well plates to study responses of a population of cells under different chemical, genetics, or radiation perturbations. Extracting quality information in bioassay development and screening is enabled by a powerful combination of multi-dye fluorescence imaging, flexible analysis algorithms and system automation. A roadblock that prevents high content screening from becoming widely used is the difficulty of processing and analyzing large amounts of image datasets generated. The challenge lies in how to convert all the images showing functions and interactions of macromolecules in live cells and tissues into quantitative numbers that can be analyzed statistically. In this report, we describe an image informatics pipeline developed for HCS, focusing on image analysis aspects of the pipeline.