Welcome to CoevDB!

What is CoevDB ?

CoevDB is a database that collects prediction of coevolution between pairs of sites within DNA coding sequences. In its current state, CoevDB contains coevolution predictions for approximately 8'200 protein families of species within the bony vertebrates (Euteleostomi clade).

A word about coevolution

The method used to predict those coevolving sites is Coev, a probabilistic model proposed by Dib et al., 2014. This evolutionary model describes the process of evolution of correlated positions within nucleic sequences based on a 16X16 substitution matrix. This markov model describes the coevolving process along a phylogenetic tree and reconstructs the ancestral states of coevolving pairs of positions. This parametric model identifies coevolving positions and estimates the profile associated with the coevolution process.

If you are not familiar with evolutionary models, a more intuitive explanation of molecular coevolution and the Coev software can be found here .

CoevDB: an unprecedented large-scale analysis

While prediction of Coevolution can be obtained using the Coev software or its companion web service, inferring coevolution on large molecular datasets while accounting for the underlying phylogenetic tree remains a computational challenge. CoevDB contains the results for closely 12 billions of coevolution analyses that required the equivalent of 650 years of computation using a single CPU (2 millions of CPU hours).

Among the billion of pairs of positions tested for coevolution, more than 3.5 millions of pairs were predicted as being potentially coevolving (Delta AIC > 25). Among these pairs, ~400'000 shows a significant support for the Coev model (Delta AIC > 40).

How to browse CoevDB ?

Results stored in CoevDB are indexed by thename of the gene analyzed as well as the pair of sites position. We offer you multiple ways to explore these results.

  • Summary statistics: on this page, all genes are displayed with their human gene name, their molecular length as well as their amount of predicted coevolving pairs. Links are made available for each gene to help you refine your search.
  • Browse all best pairs: this page enables you to select the gene of your choice and then display all its pairs of positions being potentially coevolving (Delta AIC > 25).
  • Query best pairs: this page enables you to select a gene and a position or pair of positions. The pairs of positions corresponding to this filter and being potentially coevolving (Delta AIC > 25) are then displayed.
  • Query pairs: this page enables you to select a gene and pair(s) of position. The results for all the profile of coevolution analyzed for this pair are then displayed.

See the tutorial page for more information.

Wanting a more direct access ?

If this punctual access is not sufficient for your needs, the full database is available for you to download (as MySQL dumps).

How to cite CoevDB ?

  • Link to the article in the NAR Database 2019 issue
  • Xavier Meyer, Linda Dib, Nicolas Salamin; CoevDB: a database of intramolecular coevolution among protein-coding genes of the bony vertebrates, Nucleic Acids Research, , gky986