Large Datasets in Drug Discovery: How to Collect, Summarize, and Visualize Data from Image-based Cellular Assays in Meaningful Way

• Activity Number: 111
• Type: Contributed
• Date/Time: Monday, August 9, 2004 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Graphics
• Author(s): Vadim Kutsyy*+
• Companies: Cytokinetics, Inc.
• Keywords: drug discovery ; large data ; visualize data ; cellular assays

Abstract

Automated high-content, cell-based tools are among the most powerful nonclinical methods for understanding the broad biological effects of uncharacterized compounds. High-content screening platforms include fluidics for plating and treating cells, automated microscopes for imaging, and software for image and data analysis. I will talk about image and data analysis for such screens. A typical image-based cellular assay generates more than 100,000 images, leading to multivariate data for tens of millions of individual cells. Thess data need to be presented to biologists and chemists in meaningful ways. A number of different methods and algorithms are employed to summarize data at different levels. This process creates an interesting statistical problem related to large datasets. Different tools are used to present these data to scientists and link it to images, and chemical or biological properties. This talk will describe methods and tools used by cytokinetics to facilitate analysis, visualization, and drilling down of discovery data across the enterprise.