Fig. 2: In silico sequence evolution towards functional enhancers.

a, Prediction score distribution of the sequences for the γ-KC class (n = 6,000 sequences) after each mutation. The box plots show the median (centre line), interquartile range (box limits) and 5th and 95th percentile range (whiskers) for KC-directed (blue) or random drift (orange) mutations. b, Nucleotide contribution scores for the γ-KC class of a selected random sequence in its initial form (top) and after in silico evolution (bottom). c, Prediction scores of 13 selected sequences at each mutational step. Dashed line indicates the selected iteration (10th or 15th mutation). d, In vivo enhancer activity of the cloned KC sequences with positive enhancer activity. e, Nucleotide contribution scores for the PNG class of the same selected random sequence as in b (top) and after PNG-directed mutations (bottom). f, In vivo enhancer activity of the cloned PNG sequences. Top-middle, initial random sequence; top-left, random sequence after 10 mutations toward KC evolution; top-right, random sequence after 15 mutations toward PNG evolution; bottom, three other random sequences after mutations toward PNG evolution. g, In vivo enhancer activity of the cloned genomic sequences with 6 mutations (11 for FP3). h, Nucleotide contribution scores of a selected genomic sequence in its initial form (top) and after six iterations (bottom). In b,e,h, dashed line shows the position of the mutations, the mutational order and type of nucleotide substitutions are written in between top and bottom plots and motif annotation is indicated with strong (s) or weak (w) motif instances. In d,f,g, the expected ___location of KC is shown with dashed circles. Scale bars, 100 µm.