What is the origin of the term ‘homology’? Richard Owen (1804–1892) defined homology as “the same organ under every variety of form and function”. Owen conceived of homologous structures as those that, while differing in detail, were derived from the same body plan, or ‘archetype’. By contrast, analogous structures were those that performed similar functions but did not appear to be derived from the same archetype. After Darwin, homologous morphologies were reinterpreted as having derived by divergence from a common ancestral structure. Meanwhile, analogous morphologies were thought to have arisen by convergence, such as the independent invention of wings during bird and bat evolution. So now, homology describes descent from a common evolutionary origin: two genes are homologous if they derive from the same ancestral gene. Differentiating between homology and analogy is not mere pedantry: homology allows Darwinian evolutionary theory to be applied accurately across the biosciences. And, as Theodosius Dobzhansky (1900–1975) famously remarked, “Nothing in biology makes sense except in the light of evolution”.Is sequence similarity the same as homology? Definitely not. Sequence similarity is a quantity that is agnostic of evolution. In contrast, homology is a property that describes evolutionary history. Just as with bird wings and bat wings, perceived similarities between sequences need not be due to a common evolutionary origin. Research papers sometimes wrongly quote values of ‘percent homology’. In these cases ‘percent identity’ is meant, as two genes either have a common ancestor or they do not. The only appropriate use of ‘percent homology’ is when separate portions of a gene have distinct evolutionary histories, for example as a result of a gene fusion event.How can one be sure beyond reasonable doubt that two similar sequences are homologous? Using statistics you can estimate how likely it is that randomly composed sequences yield alignment scores that are at least as high as that obtained between the real sequences in question. For example, the BLAST program reports an Expect (or E) value for each alignment (with score x), which is the number of times sequences are expected, with scores ≥x, to crop up in a search just by chance. As E gets closer to zero, the more confident one should be in a prediction of homology. Many users cautiously consider only those alignments with E-values lower than 10−3 as substantiating evidence for homology.Is any other evidence relevant? Structural similarities are important too. But once again we are faced with ‘similarities’: we cannot be sure that just because two proteins fold up in the same way it means they arose from a common ancestor. Nevertheless, spatial coincidence of active or binding sites, or unusual structure, can boost the odds of a homology prediction being correct.What about convergent evolution? As far as we can tell, the convergence of gene sequences is extremely rare. It is, by far, ‘easier’ for Nature to duplicate a gene than invent similar genes on two separate occasions. By contrast, independent invention of protein structure is often suggested to have occurred, yet for most of these cases the evolutionary provenance is unclear.What are ‘orthology’, ‘paralogy’ and ‘xenology’? These are relationships between genes best visualized in a phylogenetic tree. Orthologs are genes resulting from the splitting of different lineages – speciation. Paralogous genes arise from duplications within the same genome. Lastly, genes that have been acquired via horizontal – or ‘lateral’ – transfer between different species are referred to as xenologues.These relationships are clearly illustrated in Figure 1Figure 1. However, lineage-specific gene deletion, pseudogenisation, duplication, conversion and rapid sequence divergence can all confuse phylogenetic tree reconstruction. For example, the loss of genes A2 and B1 in Figure 1Figure 1 may cause duplication event DP1 to go undetected, and hence an erroneous assignment of paralogous genes A1 and B2 as orthologs. Gene conversion fuses sequences with contrasting heritages. It can result in a gene in one species being both orthologous and paralogous to a gene in another. Horizontal gene transfers can lead to incongruencies between gene-based and taxon-based trees which often assist the detection of xenology relationships.Figure 1The phylogenetic relationships between eight extant genes from four species denoted A, B, C and D.Three speciation (SP1, SP2, SP3), two intra-genome duplication (DP1, DP2), and one horizontal transfer (H1) events have occurred since the last common ancestor (the ‘cenancestor’) of all eight genes. The relationship of any pair of genes can be found by tracing their lines of descent upwards until they converge at the nodes: orthologs are those that join at a speciation event (an inverted ‘Y’), whereas paralogs join at a duplication event (a horizontal bar). For example, gene A1 is paralogous to A2 and B2, but orthologous to B1, C1, D1, D2 and D3; gene C1 is orthologous to all genes, except D1; and, D1 is xenologous to all other genes.View Large Image | View Hi-Res Image | Download PowerPoint SlideNote that these relationships are defined with respect to evolution, and not function. Nevertheless, they are useful in predicting function as the more recently two genes shared a common ancestor, the more likely it is that they have retained similar functions. Moreover, orthologous genes that have been spared by natural selection from deletion or duplication over many millions of years are also likely to share overlapping functions.Do the terms orthology, paralogy and xenology apply only to genes? No: the same terms can be used for genomic regions encompassing several genes, and even single nucleotide sites. For example, large chromosomal segments that arose by an intra-genome duplication are paralogous genomic regions, which some call ‘paralogons’. Similarly, sequences that have persisted essentially intact in two species since their common ancestor may be termed orthologous genomic regions.What about synteny and orthologous genomic regions? Synteny – literally the ‘same thread’ – was defined originally as relating to gene loci on the same chromosome. In comparative genomics, however, ‘synteny’ has become short-hand for ‘conserved synteny’, and used synonymously with orthologous genomic regions containing orthologous genes in a similar collinear order.Do we need new terms (neologies)? Some would say that we do. They argue that we should coin terms to describe similarities – in sequence or structure, for example – between biological molecules regardless of whether these arose by divergence from a common ancestor. Only definitions that are useful will survive, they suggest, while those that are not will be dropped (a linguistic mimicking of purifying selection). We believe that there is too much bewilderment already in the use of homology, orthology and paralogy, so introducing yet more terms appears to be asking for trouble. Moreover, the terms in current use are sufficient, when applied appropriately, to qualitatively describe the consequences of gene duplication (homologs), speciation (orthologs), intragenome duplication (paralogs) and horizontal transfer (xenologues), which are four of the major evolutionary forces acting on genes.Where can I find out more?