Send to

Choose Destination
PLoS One. 2010 Apr 1;5(4):e9989. doi: 10.1371/journal.pone.0009989.

A benchmark of parametric methods for horizontal transfers detection.

Author information

Dynamique des Structures et Interactions des Macromolécules Biologiques, Institut National de la Santé et de la Recherche Médicale UMR-S 665, Université Paris Diderot, Institut National de la Transfusion Sanguine, Paris, France.


Horizontal gene transfer (HGT) has appeared to be of importance for prokaryotic species evolution. As a consequence numerous parametric methods, using only the information embedded in the genomes, have been designed to detect HGTs. Numerous reports of incongruencies in results of the different methods applied to the same genomes were published. The use of artificial genomes in which all HGT parameters are controlled allows testing different methods in the same conditions. The results of this benchmark concerning 16 representative parametric methods showed a great variety of efficiencies. Some methods work very poorly whatever the type of HGTs and some depend on the conditions or on the metrics used. The best methods in terms of total errors were those using tetranucleotides as criterion for the window methods or those using codon usage for gene based methods and the Kullback-Leibler divergence metric. Window methods are very sensitive but less specific and detect badly lone isolated gene. On the other hand gene based methods are often very specific but lack of sensitivity. We propose using two methods in combination to get the best of each category, a gene based one for specificity and a window based one for sensitivity.

[Indexed for MEDLINE]
Free PMC Article

Supplemental Content

Full text links

Icon for Public Library of Science Icon for PubMed Central
Loading ...
Support Center