Résumé
For a long time, population geneticists have had to content themselves with analysing neutral markers to infer selection processes indirectly. The recent improvement of sequencing tools now enables them to analyse the variations of a large number of genes and therefore to spot the genes, or the amino acids, which are under the direct influence of the selection (Yang & Bielawski, 2000). In order to analyse the selection processes acting on the mutations and discriminate the different factors acting on their evolution (genetic selection or drift), it is essential to compare polymorphism data between genomes of the same species/population and divergence data between species/populations (Kimura, 1983). Recently, polymorphism data has become available for a sufficient number of genes (>10) in some model species like humans (Fay et al., 2001; Bustamante et al., 2005), the fruit fly (Bierne & Eyre-Walker, 2004) or the arabis (Bustamante et al., 2002). Some inferences have provided surprising results, suggesting that adaptation holds a more important role in the divergence between two species than the expected neutralist hypothesis (review in Eyre-Walker, 2006). However, other results suggest that the efficient size of populations could be a limitating factor for adaptative evolution (Bustamante et al., 2002), even if its role in non-neutral genetic variation is still not well understood (Gillespie, 1999; Bazin et al., 2006). Finally, the number of polymorphic sites in a given gene is often too small to make gene by gene inferences (except for some fruit fly genes) and only a global estimate can be obtained through the compiling of several genes (Bierne & Eyre-Walker, 2004). If some sequences ultra-dense in polymorphic sites existed, reliable estimates could be made at gene level.