We treat proteins as amorphous learning matter: A `gene' encodes bonds in an `amino acid' network making a `protein'. The gene is evolved until the network forms a shear band across the protein, which allows for long-range soft modes required for protein function. The evolution projects the high-dimensional sequence space onto a low-dimensional space of mechanical modes, in accord with the observed dimensional reduction between genotype and phenotype of proteins. Spectral analysis shows correspondence between localization around the shear band of both mechanical modes and sequence ripples.