## You are here

# Inconsistent distances in substitution matrices can be avoided by properly handling hydrophobic residues.

Title | Inconsistent distances in substitution matrices can be avoided by properly handling hydrophobic residues. |

Publication Type | Journal Article |

Year of Publication | 2008 |

Authors | Baussand, J, Carbone, A* |

Journal | Evol Bioinform Online |

Volume | 4 |

Pagination | 255-61 |

Date Published | 2008 |

ISSN | 1176-9343 |

Abstract | The adequacy of substitution matrices to model evolutionary relationships between amino acid sequences can be numerically evaluated by checking the mathematical property of triangle inequality for all triplets of residues. By converting substitution scores into distances, one can verify that a direct path between two amino acids is shorter than a path passing through a third amino acid in the amino acid space modeled by the matrix. If the triangle inequality is not verified, the intuition is that the evolutionary signal is not well modeled by the matrix, that the space is locally inconsistent and that the matrix construction was probably based on insufficient biological data. Previous analysis on several substitution matrices revealed that the number of triplets violating the triangle inequality increases with sequence divergence. Here, we compare matrices which are dedicated to the alignment of highly divergent proteins. The triangle inequality is tested on several classical substitution matrices as well as in a pair of "complementary" substitution matrices recording the evolutionary pressures inside and outside hydrophobic blocks in protein sequences. The analysis proves the crucial role of hydrophobic residues in substitution matrices dedicated to the alignment of distantly related proteins. |

Alternate Journal | Evol. Bioinform. Online |

PubMed ID | 19204823 |

PubMed Central ID | PMC2614183 |