bioinformatics, horizontal gene transfer, protozoa,
endosymbiotic gene transfer, Trypanosomatids
Horizontal gene transfer (HGT) is a widespread phenomenon, not only in prokaryotes. Protozoa are especially prone to HGT, as many of them live together with other organisms in close endosymbiotic relationship. Detection of horizontal gene transfer can be a difficult task, especially for transfer events from prokaryotes to eukaryotes, if species are only partly sequenced or phylogenetic relationships are not clear. Detection tools that screen complete genomes or compare gene trees with the corresponding species tree are less suitable at that point. In the reality of the HGT laboratory, decisions about the probability of gene transfer have to be made for individual genes, which may constitute an important factor in endosymbiotic relationships or specialization of pathogenicity. We recognized the demand for a more specific detection tool working on single genes and developed an algorithm combining four different but complementary approaches for HGT detection in a score-based application, i. protein domain-based information, ii. BLAST, iii. GC content and codon adaptation index, iv. multi-level trees. The multi-level tree approach avoids extensive phylogenetic analysis. We show that protein similarity can be analysed also at minimal resolution. In our new algorithm, Neighbour-Joining trees are solely used for providing an internal data structure and not a method for gold standard HGT detection by phylogenetic analysis. We substantiate the potential of the algorithm with genes from the non-human pathogenic Trypanosoma rangeli, which are compared with parasitic and symbiont-harbouring trypanosomatids to estimate the possible impact of HGT on their lifestyle.