Skip to main content

Archived Comments for: Inter-species horizontal transfer resulting in core-genome and niche-adaptive variation within Helicobacter pylori

Back to article

  1. Horizontal gene transfer across closely related bacterial strains

    Christos Ouzounis, EMBL-EBI

    13 June 2005

    Dear Editors,

    We are writing in response to the recent publication by Saunders et al. [1], where they attempt to characterize horizontal gene transfer (HGT) in the two published genomes of Helicobacter pylori strains. Typically, one might not counter such analyses, given the uncertainties surrounding this topic. However, because of the particular importance of this well-studied species as well as the breach of key principles in this field, we feel obliged to react by this short communication.

    First and foremost, none of these data are available in electronic form. Only one of the two URLs mentioned in Methods is active, providing a version of the program in executable form. Thus, we have not manually analyzed all the genes that have been reported for reasons that will be clear below. Second, the convoluted presentation of six tables with three parameter settings (di-, tri- and hexa-nucleotide compositional statistics) [1], each for the two strains of Helicobacter pylori indicates that the authors have not been able to decide which parameter value is most relevant in their analysis. A simple rank order statistic could have been useful, yet it appears that it would not show a departure from a random list of rankings across the three parameter settings. Moreover, similar work that successfully used compositional statistics for the detection of HGT has not been cited [2]; more specifically, the use of short genes (e.g. the pair JHP1321/HP1432), listed by Saunders et al. [1] in all tables should have been excluded [2]. Third, certain conceptual issues related to HGT analysis have not been addressed, in particular the strain specificity of the findings, the extent to which these genes are present in other species and independent experimental corroboration of the results from a variety of sources [3].

    Recently, methods that infer HGT on the basis of either compositional statistics [2] or phylogenetic distribution [4] have been developed. The basic ideas are that genes must differ in GC content and codon usage from their own resident genome background [2] and possibly also be phylogenetically interspersed across different species [4], to be strong candidates for HGT. None of these two fundamental principles have been examined in the analysis under consideration [1], not even as alternative control hypotheses. In fact, the use of sliding window-based tri-nucleotide statistics without reference to codon preference in protein-coding genes is simply absurd. Furthermore, the phylogenetic distribution of most of the reported genes is canonical and does not point to any anomalies with regard to distorted species or sequence similarity patterns (e.g. a H. pylori gene that would be highly similar to a human gene). The claims that e.g. FtsK, XerD, MutS and PolA are horizontally transferred in H. pylori [1] cannot be supported by any other facts known to date. In fact, MutS is detectable in ca. 100 species and PolA in all genomes, without a peculiar phylogenetic scattering. These arguments suggest that most of the identified H.pylori genes [1] have not been acquired via HGT and they are simply false positives.

    We have previously conducted an in-depth study of this species, with the aim to identify (i) strain-specific genes [5] and (ii) genes acquired by HGT [6]. We have identified 162 strain-specific genes, which might be considered as prime candidates for HGT. Of these, only 78 genes are amenable to compositional analysis [6] of which 77 are independently confirmed by experimental analysis as being strain-specific [7]. This remains, to our knowledge, the best reference set to examine HGT between two highly similar bacterial strains. We have shown that 60 of these genes (77%) do not have homologues in more than three bacterial lineages and only three have a homologue in the related Campylobacter jejuni genome. Also, 69 of these genes (88.5%) are outliers in terms of composition. On the basis of these numbers, our false-positive rate is estimated at 1.3% (1 out of 78, not found as strain-specific by experiment) and our false-negative rate is estimated at 11.5% (9 out of 78), with regard to HGT (but not strain specificity). If this set is considered as the gold-standard, one can only wonder what are the corresponding false detection rates (for example, only 19 strain-specific genes are identified for any parameter setting, pointing to a false-negative rate of 88% and an even higher false-positive rate).

    It is imperative for computational genomics to follow a scientifically consistent path forward and not simply invent unusable methods that create confusion – especially for experimental biologists, without losing credibility and a view for the future.

    Sincerely,

    PJ Janssen • Belgian Nuclear Research Centre, Mol, Belgium

    CA Ouzounis • European Bioinformatics Institute, Cambridge, UK

    References

    1. NJ Saunders, P Boonmee, JF Peden, SA Jarvis: Inter-species horizontal transfer resulting in core-genome and niche-adaptive variation within Helicobacter pylori. BMC Genomics 2005, 6:9.

    2. S Garcia-Vallve, A Romeu, J Palau: Horizontal gene transfer in bacterial and archaeal complete genomes. Genome Res 2000, 10:1719-25.

    3. EA Joyce, K Chan, NR Salama, S Falkow: Redefining bacterial populations: a post-genomic reformation. Nat Rev Genet 2002, 3:462-73.

    4. V Kunin, CA Ouzounis: The balance of driving forces during genome evolution in prokaryotes. Genome Res 2003, 13:1589-94.

    5. PJ Janssen, B Audit, CA Ouzounis: Strain-specific genes of Helicobacter pylori: distribution, function and dynamics. Nucleic Acids Res 2001, 29:4395-404.

    6. S Garcia-Vallve, PJ Janssen, CA Ouzounis: Genetic variation between Helicobacter pylori strains: gene acquisition or loss? Trends Microbiol 2002, 10:445-7.

    7. N Salama, K Guillemin, TK McDaniel, G Sherlock, L Tompkins, S Falkow: A whole-genome microarray reveals genetic diversity among Helicobacter pylori strains. Proc Natl Acad Sci U S A 2000, 97:14668-73.

    Competing interests

    No competing interests.

Advertisement