Finding Genes

Abstract

The definition of the term gene changed over time and based on new findings. Nowadays, genes are broadly defined as a union of genomic sequences encoding a coherent set of potentially overlapping functional products. Gene numbers between species are highly variable and gene loss and gain are common. The fate of most duplicated genes is pseudogenization (nonfunctionalization). However, duplicated genes might be retained if new functions are evolved in one of the copies (neofunctionalization) or if different ancestral functions are retained in different copies (subfunctionalization). Homologous genes are derived from a common ancestor and identified through similarity. Copies which arose due to speciation events are called orthologs, whereas those that arose by duplication events are paralogs. In the ideal case, phylogenetic reconstructions based on orthologs coincide with the species tree. Orthology inference methods are classified into graph-based methods and tree-based methods. Graph-based methods usually rely on similarity scores derived from pairwise alignments, whereas tree-based methods use phylogenetic analyses. Several databases comprising large sets of gene orthology predictions are available. With the help of hidden Markov models (HMMs), predefined sets of orthologs can be retrieved from genomic or transcriptomic datasets. The PFAM database hosts a large collection of protein families predicted from HMMs. Orthology is further used for the inference of gene (or gene product) function). Based on the ortholog conjecture, orthologs are more likely to indicate conserved function than paralogs. This assumption underlies most current genome annotations. The Gene Ontology (GO) project cataloged functional predictions of gene products using a controlled vocabulary for three different classes: cellular component, biological processes and molecular function.