Skip to main content

Table 3 Predictions by genetic algorithms for the categorisation of different papillomaviruses according to the presence/absence of TFBS in their URR.

From: Different papillomaviruses have different repertoires of transcription factor binding sites: convergence and divergence in the upstream regulatory region

input

grouping

correctly predicted

erroneously predicted

unable to predict

all viruses

6 categories, random

44%

66%

0%

 

6 categories, taxa: alpha beta gamma+xi kappa+lambda+mu delta+zeta theta+iota

64% (α 85%) (β 84%)

18 (α 7%) (β 8%)

22% (α 7%) (β 8%)

 

2 categories, random

61%

39%

-%

 

2 categories primate/non primate

90%

2%

8%

human PVs

4 categories, random

30%

56%

14%

 

4 categories, taxa: alpha beta gamma mu

76% (α 79%) (β 92%)

12% (α 85) (β 0%)

12% (α 12%) (β 8%)

alpha + beta PVs

2 categories, random

39%

39%

22%

 

2 categories, taxa: alpha beta

97%

3%

0%

  1. Genetic algorithms were trained with the matrix describing the presence/absence of TFBS in the URR of the papillomaviruses. An additional column was provided, including either real information about the pertenence of each virus to a genus, or about the nature of the infected host. As a control, additional training was performed on the same matrix providing a column with a random categorisation. Numbers in the table reflect the percentage of viruses for which the genetic algorithms rendered correct predictions, false predictions, or could not formulate any prediction, respectively. Additional values in brackets show the specific accuracies of the predictions for the alpha and beta papillomaviruses. Genetic algorithms were able to discern the pertenence of a given virus to a genus, attending only to the repertoire of TFBS in the URR. Specifically, alpha and beta papillomaviruses were recognised as separate definite clades with high confidence in most of cases. The main targets of both genera are basal cells in the stratified squamous epithelia. These results suggest that alpha and beta papillomaviruses take advantage of different control mechanisms in the same target cell, and/or that they infect different subsets of cells, which are histologically indistiguishable. Finally, genetic algorithms were able to recognise whether a given papillomavirus infects or not primates, regarding only the TFBS patterns in the URR. Papillomaviruses infecting primates are phylogenetically distant, and separated long before the appearance of the primates themselves. Therefore their clustering together according to the repertoire of TFBS suggests a uniformity of regulatory mechanisms, achieved by convergent evolution.
Back to article page

BMC Ecology and Evolution

ISSN: 2730-7182

Contact us