It is well known among molecular biologists that proteins with a common ancestor and that perform the same function in similar organisms, can have rather different amino-acid sequences. Mutations have altered the amino-acid sequences without affecting the function. A simple model of a protein in which the interactions are encoded by sequences of bits is introduced, and used to study how mutations can change these bits, and hence the interactions, while maintaining the stability of the protein solution. This stability is a simple minimal requirement on our model proteins which mimics part of the requirement on a real protein to be functional. The properties of our model protein, such as its second virial coefficient, are found to vary significantly from one model protein to another. It is suggested that this may also be the case for real proteins in vivo.