Fig. 14
Phase of rhythmicity prediction. Prediction of the phase of rhythmicity in selected baboon tissues/organs from a set of transcripts in a human blood sample. Schematic diagram of the development and validation of a multivariate whole-blood mRNA-based predictor of tissue/organ-specific rhythmicity using a Partial Least Squares Regression (PLSR) model. Input to the model is the human blood expression X of g candidate prognostic genes in n training samples. The model’s target, Y, is the human blood sampling time relative to the acrophase of the tissue-marker gene. PLSR is used for model selection (number of latent factors and predictor genes p) via cross-validation and to fit the Ŷ =B X + Bo model. Validation of the model is performed by comparing the predicted vs the ‘observed’ acrophase of the tissue-marker (blood sampling time – Ŷ) using an independent set of validation samples