Martin Macko. Algoritmus pre hľadanie vzdialených funkčných ortológov a jeho aplikácia na OB-fold doménu. PhD thesis, Comenius University in Bratislava, 2013. Supervised by Tomáš Vinař.
Download preprint: 13mackoth.pdf, 1248Kb
Download from publisher: http://alis.uniba.sk/storage/dpg/dostupne/FM/2013/2013-FM-40974
Related web page: not available
Bibliography entry: BibTeX
Identification of evolutionarily related proteins (orthologs) is an important step towards understanding protein function in living organisms. This problem is difficult in many cases because distantly related proteins have often too divergent amino-acid sequences, so this relation may not be identified by traditional methods based mainly on sequence similarity. In these cases search should use combination of structural and sequential features of protein. In this thesis, we propose a new approach to the search for remote functional orthologs, where important features of protein sequence and structure are represented by a descriptor created by human expert. This descriptor can be used to search for described features in candidate protein sequences. For the problem of ortholog identification with descriptor, we develop scoring scheme which combines sequence profiles and support vector machines to evaluate alignment of the descriptor to the candidate sequence, and we develop algorithms which that the best possible alignment. We demonstrate our approach on the example of telomere-binding OB-fold domain. Our method can distinguish between Telo_bind family members and negatives, and also identifies proteins containing a related OB-fold domain.