How is structural divergence related to evolutionary information?

ALEXANDER MONZÓN; CRISTINA MARINO; DIEGO JAVIER ZEA; CRISTINA MARINO-BUSLJE; GUSTAVO PARISI

Buenos Aires

Congreso; International Society for computational biology/Asoc. Arg de Bioinformática y Biol. Computacional; 2016

BackgroundMultiple Sequence Alignments (MSA) populated with distant homologous sequences are commonlyused to derive evolutionary information. Furthermore, they are use to link sequence with structuralinformation. It is also usual to use conservation to get an intuition of the evolutionary rate of a siteand to use covariation between sites to get an understanding of coevolution.Several methods to measure conservation and covariation are based on the information theory. Inparticular, the Shannon entropy and the Kullback-Leibler divergence are commonly used as ameasure of variation and conservation, respectively, of a protein site (Johansson and Toh 201 0).Mutual Information (MI ) was widely used as a measure of covariation between protein residues. Ithas been shown that mutual information scores perform better at protein contact prediction withMSAs populated with a high number of diverse sequences (Buslje et al. 2009; Dunn, Wahl, and Gloor2008). Since Pfam alignments meet these conditions, they are commonly used to calculate thisscores.Also, it is usually assumed that the key structural determinant of site-specific rates of evolution isthe relative solvent accessible surface area (RASA) (Echave, Spielman, and Wilke 201 6). Previousworks has linked evolutionary and structural information using a single structure from the MSA (Yehet al. 201 4; Marino Buslje et al. 201 0).Lesk and Chothia (Chothia and Lesk 1 986) have shown that the overall extent of structural change isdirectly related to the extent of the sequence divergence between two proteins. We can expectsignificant structural variation between different proteins in an MSA.Here, we examined the extent of variation between different structures present in an MSA, anddetermined that there is a mean change of RASA per residue of 26.80%. Also, on average, the42.78% of the MSA columns change its exposed/buried status, and 54.95% of the contacts betweenresidues change. Furthermore, there is a maximum root mean square deviation (RMSD) of 4.41 Åbetween different structures in an MSA on average.Here we demonstrate how it is possible to lead to different results when comparing structural withevolutionary information when ignoring structural divergence