An Information Ratio Criterion for Integration of High-Throughput Data Sources with Applications to Quality Control and Novelty-Detection in Pluripotent Stem Cell Cultures

• Activity Number: 464
• Type: Contributed
• Date/Time: Wednesday, August 3, 2011 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Bernhard Michael Schuldt*+ and Michael Lenz and Johanna Goldmann and Franz-Josef Müller and Jeanne Loring and Andreas Schuppert
• Companies: AICES Graduate School and AICES Graduate School and Freie Universität Berlin and Zentrum für Integrative Psychiatrie and The Scripps Research Institute and AICES Graduate School
• Keywords: microarray ; DNA-Methylation ; next-generation sequencing ; novelty detection ; quality control

Abstract

Human pluripotent stem cells have become a focus of academic and industry researchers alike. Currently, the quality of such in vitro preparations is assessed mainly with qualitative, low-content assays such as the teratoma assay. Alternatively high-thoughput methods can be used to generate data-driven models to evaluate pluripotency, to assess quality and to detect novel features with biological implications in the data . Technologies such as mRNA-, SNP- and methylation arrays as well as next-generation deep sequencing have been proposed for this task. To establish an integrative quality control framework in a fast developing field it is key to understand how information is preserved with new profiling technologies. We use the information ratio (IR), a measure that compares gene-wise measurements to large scale patterns, to develop a classification of various quality control and novelty detection tasks. We evaluated the IR heuristic on a large, heterogeneous stem cell data set and developed practical guidelines for an integrated quality control workflow that can be easily extrapolated to data generated in other laboratories or can be applied to other fields.