Dating with transfers

Adrián Davín (1) , Gergely J Szöllosi (2) , Eric Tannier (2, 3) , Bastien Boussau (2) , Vincent Daubin (2)

1 LBBE - Laboratoire de Biométrie et Biologie Evolutive - UMR 5558
2 BPGE - Bioinformatique, phylogénie et génomique évolutive
3 BEAGLE - Artificial Evolution and Computational Biology

To reconstruct the timing of the diversification of life on Earth, biologists combine fossil evidence with inferences drawn from the comparison of genome sequences. These inferences are based on the molecular clock or on a softer version of it, the relaxed molecular clock. This approach consists of estimating the divergence between sequences and then, assuming that mutations occur clockwise , trying to determine the age of the ancestral sequence. This method can be refined by using diverse models that relax the hypothesis of constant pace of evolution and consider that sequences can evolve at different speeds. Some models assume that these rates of evolution are independent along the different branches of the tree relating the different sequences of DNA; some others consider that the rates are correlated among related branches, so the rate of a given branch is inherited to some extent from the parental one. Which is of these methods perform best is still heavily debated. In spite of the sophistication of the different models, calculating these rates is not a trivial problem and the best estimates of divergence time have usually very wide confidence intervals. To overcome this problem scientists can use fossils, that can be independently dated using methods such as stratigraphy or radiometry. Fossils are useful because they provide external information that can be used to constrain the positions of the nodes in a species tree, improving the accuracy of the estimates of the molecular clock. Combining relaxed molecular clock estimates and fossil is in active field of research in phylogenetics [1]. However, fossils are extremely scarce in the geological record. For about 80 % of the history of life, all organisms were unicellular, which means that finding fossils becomes an almost impossible task. Bones and hard shells are easy to be preserved but they just became frequent after the Cambrian explosion, when all the major animal clades appear at sudden. Before that Earth was dominated by bacteria and to a minor extent, small eukaryotes. These organisms are extremely small organisms with no hard parts that can fossilize easily. On top of that, for the few existent fossils we have there is very little certainty about the clades to which they belong, since morphological features cannot be used to place them in a phylogenetic tree. This means that if we are interested in studying what happened in the distant past, we have very little help coming from fossils and we must rely almost exclusively in the information conveyed by the DNA. As we previously stated, this is a hard problem since the estimates of the molecular clock can vary widely. We need accurate calculations if we want to know for example when did Eukaryotes diversify or when did cyanobacteria appear on Earth. To overcome these problems, we propose a new method of dating, based on the DNA sequence complementary to the molecular clock. Lateral gene transfer (LGT) is a common and almost universal phenomenon in nature, where different species (sometimes even species belonging to different domains) exchange genes. This can be detected using differences between species trees and gene trees. We do this using ALE, a method to reconcile species trees and gene trees that allows.

Domaines: Bio-informatique [q-bio.QM]

Adrián Davín, Gergely J Szöllosi, Eric Tannier, Bastien Boussau, Vincent Daubin. Dating with transfers. Journées Ouvertes Biologie Informatique Mathématiques, 2016, Lyon, France. ⟨hal-01394410⟩