Fig. 6: A methylome-based predictive model captures both cellular and spatial characteristics of neurons.
From: DNA methylation atlas of the mouse brain at single-cell resolution
a, Schematic of the model that predicts both cell-type identity and spatial origin. b, c, Model performance on the prediction of cell subtypes (b), and dissection regions (c). Per cent accuracy is shown. d, Examples of using the model to predict cell spatial origin (maximum prediction probability in parentheses). e, Evaluation of importance of features (principal components) for spatial origin prediction accuracy. ‘Acc. Δ%’ denotes the average prediction accuracy decrease percentage. f, The Foxp2 gene body mCH fraction in each cortical dissection region group.
