4. Site-specific amino-acid replacement matrices from structurally constraind protein evolution simulations

FORNASARI, MS, PARISI, GUSTAVO AND ECHAVE J

MOLECULAR BIOLOGY AND EVOLUTION

Año: 2002 vol. 19 p. 352 - 356

0737-4038

Evolutionary amino acid replacement rates dependon local structural environment (Overington et al. 1992;Koshi and Goldstein 1995). Recent models of proteinevolution aim to take such site heterogeneity into accountby using site-specific amino acid replacement matrices(Lio and Goldman 1998; Thorne 2000). This is adifficult task, mainly because of a lack-of-data problem:too many sequences would be needed to fit the largenumber of parameters required to specify the replacementmatrix for each site. Two strategies have been usedto tackle this problem. A first type of model assumesthat protein sites can be classified into a limited numberof structural classes. Then, class-specific replacementmatrices are obtained by analyzing large databases ofsequences (Thorne, Goldman, and Jones 1996; Lio et al.1998). These models are promising, provided the assumptionof universality of structural classes is a reasonableone. However, such replacement matrices cannotbe fitted to the actual sequence family under analysis.A second strategy is to develop models that dependon a reduced number of parameters (Halpern and Bruno1998; Koshi and Goldstein 1998). These models havethe advantage that they can, in principle, be fitted to theanalyzed family. On the other hand, it is possible thatthe high reduction in the number of parameters oversimplifiessuch models to be generally applicable. Here wereport a third strategy, which allows the determinationof site-specific replacement matrices adapted to the sequencefamily under study, using just one member ofthe known structure. The lack-of-data problem is overcomeby using large amounts of simulated sequence datagenerated using a structurally constrained protein evolution(SCPE) model developed recently (Parisi andEchave 2001). We apply the method to a case taken asan example and compare it with other models of proteinevolution.