Statistical mechanics model of protein folding : short and long chains have different folding transitions

A statistical mechanics model of a polypeptide chain is used to simulate the folding process. Each residue is defined by a set of independent characters within a given sequence. Simulations of this model in a mean-field (infinite-dimensional) geometry show that (i) short chains do not fold, (ii) medium chains fold according to a spin-glass-like «transition» (where most characters must be partially satisfied) and (iii) long chains fold according to a Mattis-like «transition» (where only one dominant character is almost completely satisfied). This change in folding mechanism associated to chain length may be relevant to the existence of multidomain proteins Nous considerons un modele de mecanique statistique pour simuler le repliement d'une chaine polypeptidique. Chaque residu de la chaine est defini par un ensemble de caracteres independants. Les simulations Monte Carlo pour une geometrie de champ moyen (dimension infinie) montrent que (i) les chaines courtes ne se replient pas, (ii) la transition de repliement des chaines de longueur intermediaire est du type «verre de spins» (ou tous les caracteres doivent etre partiellement satisfaits), (iii) la transition de repliement des chaines longues est du type «Mattis» (ou un seul caractere dominant est presque completement satisfait). Ce modele est peut-etre applicable au cas de proteines multidomaines