Evolution of human tissue-specific genes. Rates, expression and interactions

ARBIZA L., MINGUEZ, P., DOPAZO J. AND H. DOPAZO

Valencia, España

Jornada; VIII Jornadas de Bioinformática; 2008

Red Temática Nacional de Bioinformática

Evolution of Human­Tisue Specific Genes. Rates, Expression and Interactions Leonardo Arbiza, Pablo Minguez, Joaquín Dopazo and Hernán Dopazo Bioinformatics Department. Centro de Investigación Príncipe Felipe Valencia. Spain Several recent studies have reached different conclusions regarding positive selection on brain specific genes along the evolutionary lineage leading to modern humans. While in one case, an acceleration of the rate of substitution in brain­specific genes of primates compared to that of rodents was interpreted as evidence for the action of positive selection among these genes, others have found very little evidence for positive selection among brain specific genes in the human lineage. To resolve this apparent disparity, we present a re­analysis of genomic data from human, chimpanzee, mouse, and rat, in addition to other previously published data sets using maximum likelihood methods and alternative methods for measuring and combining estimates of dN/dS. Additional factors, such as interactomic, expression level, and expression breadth data associated to brain and other tissue specific genes are also examined. Measurements of evolutionary rates were conducted using yn00 and CodeML (ML pairwise, branch, and branch site models) from the PAML package . Overall estimates of the value of ω from all genes in a category were estimated as the standard error (SE) weighted average. Expression data was obtained from both the human­specific Affymetrix MAS5 Absent/Present calls for the HG­U133A dataset from Genomics Institute of the Novartis Research Foundation (GNF, avilable at http://wombat.gnf.org/) and the binary tissue expression pattern search of Genenote from the Weizmann Institute of Science. Datasets were analyzed separetely. Tissue specific genes were thus defined in two different manners for each dataset. For the GNF dataset, tissues were grouped by organ and function, and tissue specificity was asigned to genes present in any number of tissues belonging to only one of the resulting groups. For the Genenote dataset, genes overexpressed in one of the tissues and under expressed in all of the rest were selected. In both cases tissues associated to disease were excluded. Ortholog annotations from human, chimp, mouse, and rat and the corresponding coding sequences the largest transcript for each gene (CDS) were retrieved from the Ensembl v42. CDSs were aligned with Muscle using translated protein sequences as templates. Data from 5 of the most complete interactomic databases available (HPRD,BIND,DIP,MINT,IntAct) were cooerced to obtain protein protein interactions confirmed in at least two of the databases in order to compare the connectivity and betweenness among tissue specific gene groups. Methods used to combine dN/dS estimates, the estimator of dN/dS, and the particularities of the genes included in the analysis, influence conclusions regarding comparisons of rates of evolution among different categories of genes. We find that although variations exist among different datasets, a consistent lack of evidence for a higher dN/dS ratio in human brain specific genes is observed. Brain is in fact the catergory with the lowest rates while the highest rates belong to liver, pancreas, and prostate or testis specific genes. Moreover, we found that tissues show interaction parameters not different from those of random groups except brain on limit of higher than random in connections. Lineage specific branch­site tests on tissue­specific genes do not show more evidence for positive selection in the human than in the chimpanzee lineage.