Additional file 1: of A purely bioinformatic pipeline for the prediction of mammalian odorant receptor gene enhancers

Figure S1. K-means clustering produces an alternative architecture for the mouse main olfactory epithelium (MOE) list. A. Single-chromosome line charts reporting (in dark cyan) the ratio between the value of the Bayesian information criterion (BIC) and the number of genes (n), for each imposed number of loci (k). k* indicates (in black) the ideal k value for k-means clustering; k~ indicates (in red) the number of loci found, for the same chromosome, by our distance-based clustering method (for threshold = 1 Mb). Small boxes (indicated by black arrows) magnify graph areas around k* and k~. B. Chromosome charts for the mouse MOE list, using a 1 Mb cutoff (left) or k-means clustering (right). Distance-based loci are reported as magenta intervals (for clusters) or green squares (for solitary genes); solitary genes are annotated on their sense strand (be it plus, +, or minus, -). k-means-based loci are invariably reported as dark cyan intervals. A location containing oversplit clusters is magnified (black shadowed box). Chromosome bands represent Giemsa staining. (TIFF 7829 kb)