A deep learning and novelty detection framework for rapid phenotyping in high-content screening

• Published in: Molecular biology of the cell Vol. 28; no. 23; pp. 3428 - 3436
• Main Authors: Sommer, Christoph, Hoefler, Rudolf, Samwer, Matthias, Gerlich, Daniel W
• Format: Journal Article
• Language: English
• Published: United States The American Society for Cell Biology 07.11.2017
• Subjects: Algorithms; Animals; Biological Variation, Population - genetics; High-Throughput Screening Assays - methods; High-Throughput Screening Assays - statistics & numerical data; Humans; Machine Learning; Numerical Analysis, Computer-Assisted; Phenotype; Software; Statistics as Topic - methods
• ISSN: 1059-1524; 1939-4586
• DOI: 10.1091/mbc.e17-05-0333

Supervised machine learning is a powerful and widely used method for analyzing high-content screening data. Despite its accuracy, efficiency, and versatility, supervised machine learning has drawbacks, most notably its dependence on a priori knowledge of expected phenotypes and time-consuming classifier training. We provide a solution to these limitations with , a generic novelty detection and deep learning framework. Application to several large-scale screening data sets on nuclear and mitotic cell morphologies demonstrates that enables discovery of rare phenotypes without user training, which has broad implications for improved assay development in high-content screening.