Fig 1. Screening strategy diagram.

IC data sets were collected from five centres (WTSI, ICS, TCP, RBRC and HMGU). Baseline data from >2000 wild-type C57BL/6N mice were used to determine parameters for identification of circadian misalignment by visual assessment and developing a machine learning. Discovery of phenodeviance was achieved by screening datasets from 750 mutant lines against primary criteria (mean data) and secondary criteria (at least 50% similar phenotypes, effect size > 1.2 and p-value < 0.001). Validation was conducted by further light entrainment experiments using mutant mice. WTSI: Wellcome Trust Sanger Institute; ICS: Institut Clinique de la Souris; TCP: The Centre for Phenogenomics; RBRC: Riken Bio-Resource Center; HMGU: German Mouse Clinic (Helmholtz Zentrum München). The light schedules in the procedure room were labelled as indicated.