An algorithm for the identification of similar oligopeptides between amino acid sequences

Molecular mimicry is the origin of common structural patterns in sequences of viral and host proteins, and it appears to be related to the development of autoimmune diseases. The identification of structural molecular similarities among viral and host proteins is thus very relevant in the development of engineered antiviral vaccines to avoid potentially dangerous effects. In this respect identifying pairs of similar oligopeptides between given proteins, independently of the overall degree of similarity of their amino acid sequences, is of interest. To this aim we have designed and implemented an algorithm capable of finding and classifying (with respect to their statistical significance) all possible pairs of similar oligopeptides between two proteins irrespective of length, number, location and ordering of the pairs along the sequences. The algorithm is very efficient and much more suited for this kind of local search than standard alignment programs. The latter, dealing with the sequences as a whole, are, in these cases, of very limited applicability. We have used the algorithm to compare a glycoprotein of the human immunodeficiency virus (HIV) type 1 and with the beta-chains of human leukocyte antigen (HLA). Besides a previously identified peptide, we have found a new peptide located in the fusion site of HIV that shares high similarity with the transmembrane domains of HLA.