Skip to Main content Skip to Navigation
Journal articles

Multiple sequence comparison - A peptide matching approach

Abstract : We present in this paper a peptide matching approach to the multiple comparison of a set of protein sequences. This approach consists in looking for all the words that are common to q of these sequences, where q is a parameter. The comparison between words is done by using as reference an object called a model, In the case of proteins, a model is a product of subsets of the alphabet Sigma of the amino acids. These subsets belong to a cover of Sigma, that is, their union covers all of Sigma. A word is said to be an instance of a model if it belongs to the model, A further flexibility is introduced in the comparison by allowing for up to e errors in the comparison between a word and a model. These errors may concern gaps or substitutions not allowed by the cover. A word is said to be this time an occurrence of a model if the Levenshtein distance between it and an instance of the model is inferior or equal to e. This corresponds to what we call a Set-Levenshtein distance between the occurrences and the model itself. Two words are said to be similar if there is at least one model of which both are occurrences. In the special case where e = 0, the occurrences of a model are simply its instances, If a model M has occurrences in at least q of the sequences of the set, M is said to occur in the set. The algorithm presented here is an efficient and exact way of looking for all the models, of a fixed length k or of the greatest possible length k(max), that occur in a set of sequences. It is linear in the total length n of the sequences and proportional to (e + 2)(2e + 1)k(e + 1)p(e + 1)g(k) where k much less than n is a small value in all practical situations, p is the number of sets in the cover and g is related to the latter's degree of nontransitivity. Models are closely related to what is called a consensus in the biocomputing area, and covers are a good way of representing complex relationships between the amino acids.
Document type :
Journal articles
Complete list of metadatas
Contributor : Marie-France Sagot <>
Submitted on : Thursday, May 3, 2012 - 11:57:04 PM
Last modification on : Friday, January 8, 2021 - 11:22:05 AM

Links full text




Marie-France Sagot, Alain Viari, H Soldano. Multiple sequence comparison - A peptide matching approach. Theoretical Computer Science, Elsevier, 1997, 180 (1-2), pp.115--137. ⟨10.1016/S0304-3975(96)00137-5⟩. ⟨hal-00694274⟩



Record views