Limits...
Characterizing gene family evolution.

Liberles DA, Dittmar K - Biol Proced Online (2008)

Bottom Line: However, they are constructed in different manners, their data analyzed and interpreted differently, with different underlying assumptions, leading to sometimes divergent conclusions.Lastly, we question the utility of layers of homology and the meaning of homology at the character state level in the context of sequence evolution.From this, we move forward to present an idealized strategy for characterizing gene family evolution for both systematic and functional purposes, including recent methodological improvements.

View Article: PubMed Central - HTML - PubMed

Affiliation: Graduate School of Biomedical Sciences, UMDNJ. liberles@uwyo.edu

ABSTRACT
Gene families are widely used in comparative genomics, molecular evolution, and in systematics. However, they are constructed in different manners, their data analyzed and interpreted differently, with different underlying assumptions, leading to sometimes divergent conclusions. In systematics, concepts like monophyly and the dichotomy between homoplasy and homology have been central to the analysis of phylogenies. We critique the traditional use of such concepts as applied to gene families and give examples of incorrect inferences they may lead to. Operational definitions that have emerged within functional genomics are contrasted with the common formal definitions derived from systematics. Lastly, we question the utility of layers of homology and the meaning of homology at the character state level in the context of sequence evolution. From this, we move forward to present an idealized strategy for characterizing gene family evolution for both systematic and functional purposes, including recent methodological improvements.

No MeSH data available.


An evolutionary trajectory of homologous sites leading to parallelevolution and to divergent followed by convergent evolution, bothgenerating homoplasy, is shown. Such a substitution pattern is notparticularly improbable under many models of sequence evolution and canreadily be found across gene families. The resulting alignmentscorresponding to homology and the non-homologous alternative are shownbelow. No standard multiple sequence alignment program will produce thealignment indicative of non-homology and this alignment is not reflective ofthe evolutionary history of the character. However, the non-homologoustreatment is the logical conclusion of considering homoplasious sites to benonhomologous.
© Copyright Policy - open acces
Related In: Results  -  Collection


getmorefigures.php?uid=PMC2683547&req=5

Figure 2: An evolutionary trajectory of homologous sites leading to parallelevolution and to divergent followed by convergent evolution, bothgenerating homoplasy, is shown. Such a substitution pattern is notparticularly improbable under many models of sequence evolution and canreadily be found across gene families. The resulting alignmentscorresponding to homology and the non-homologous alternative are shownbelow. No standard multiple sequence alignment program will produce thealignment indicative of non-homology and this alignment is not reflective ofthe evolutionary history of the character. However, the non-homologoustreatment is the logical conclusion of considering homoplasious sites to benonhomologous.

Mentions: Further, from genomic data, the distinction betweenhomology and homoplasy is artificial, as homoplasy canbe observed for homologous characters. Thus, as shown inFig. 2, a clear case of common ancestry (and thushomology) can be made for the following evolutionarytrajectories showing homoplasy. The first nucleotideposition underwent parallel evolution and the secondinvolved divergent evolution followed by convergentevolution. At the amino acid level, and especially at theDNA level, numerous characters showing these patternsof evolution can be found involving closely relatedspecies. Such patterns are not unexpected for homologoussites, given vertebrate substitution rates and the largenumber of sites in vertebrate genomes (e.g. TAED (7);HOVERGEN (9)). Further, the molecular procedure basedupon the traditional (morphological) homologyperception would require homoplasious sites to be placedin different (separate) columns in a multiple sequencealignment, as indicated in the Fig. 3 (20). In fact, this is notcommon practice for homoplasious sites in molecular data, as evolutionary models have been used tocharacterize the process of insertion and deletion, themost common process generating non-homologouspositions in an a priori alignment. The evolutionaryassessment of insertion and deletion can be done inconjunction with sequence evolution to determine if ahomoplasious site is more likely to not be homologousand have arisen by insertion and/or deletion (nonhomologoushomoplasy) or if the homoplasious site islikely to actually be homologous through shared commonancestry (homologous homoplasy) (27-28). Further,likelihood models are driving this process forward with adefinitive statement of homology through simultaneous oriterative alignment and phylogenetic tree calculation todifferentiate in a model-based way between homologoushomoplasy and non-homologous homoplasy, whileconsidering evolutionary information from gaps (indels)(29-30). A similar iterative inference can also be madeusing parsimony (31).


Characterizing gene family evolution.

Liberles DA, Dittmar K - Biol Proced Online (2008)

An evolutionary trajectory of homologous sites leading to parallelevolution and to divergent followed by convergent evolution, bothgenerating homoplasy, is shown. Such a substitution pattern is notparticularly improbable under many models of sequence evolution and canreadily be found across gene families. The resulting alignmentscorresponding to homology and the non-homologous alternative are shownbelow. No standard multiple sequence alignment program will produce thealignment indicative of non-homology and this alignment is not reflective ofthe evolutionary history of the character. However, the non-homologoustreatment is the logical conclusion of considering homoplasious sites to benonhomologous.
© Copyright Policy - open acces
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC2683547&req=5

Figure 2: An evolutionary trajectory of homologous sites leading to parallelevolution and to divergent followed by convergent evolution, bothgenerating homoplasy, is shown. Such a substitution pattern is notparticularly improbable under many models of sequence evolution and canreadily be found across gene families. The resulting alignmentscorresponding to homology and the non-homologous alternative are shownbelow. No standard multiple sequence alignment program will produce thealignment indicative of non-homology and this alignment is not reflective ofthe evolutionary history of the character. However, the non-homologoustreatment is the logical conclusion of considering homoplasious sites to benonhomologous.
Mentions: Further, from genomic data, the distinction betweenhomology and homoplasy is artificial, as homoplasy canbe observed for homologous characters. Thus, as shown inFig. 2, a clear case of common ancestry (and thushomology) can be made for the following evolutionarytrajectories showing homoplasy. The first nucleotideposition underwent parallel evolution and the secondinvolved divergent evolution followed by convergentevolution. At the amino acid level, and especially at theDNA level, numerous characters showing these patternsof evolution can be found involving closely relatedspecies. Such patterns are not unexpected for homologoussites, given vertebrate substitution rates and the largenumber of sites in vertebrate genomes (e.g. TAED (7);HOVERGEN (9)). Further, the molecular procedure basedupon the traditional (morphological) homologyperception would require homoplasious sites to be placedin different (separate) columns in a multiple sequencealignment, as indicated in the Fig. 3 (20). In fact, this is notcommon practice for homoplasious sites in molecular data, as evolutionary models have been used tocharacterize the process of insertion and deletion, themost common process generating non-homologouspositions in an a priori alignment. The evolutionaryassessment of insertion and deletion can be done inconjunction with sequence evolution to determine if ahomoplasious site is more likely to not be homologousand have arisen by insertion and/or deletion (nonhomologoushomoplasy) or if the homoplasious site islikely to actually be homologous through shared commonancestry (homologous homoplasy) (27-28). Further,likelihood models are driving this process forward with adefinitive statement of homology through simultaneous oriterative alignment and phylogenetic tree calculation todifferentiate in a model-based way between homologoushomoplasy and non-homologous homoplasy, whileconsidering evolutionary information from gaps (indels)(29-30). A similar iterative inference can also be madeusing parsimony (31).

Bottom Line: However, they are constructed in different manners, their data analyzed and interpreted differently, with different underlying assumptions, leading to sometimes divergent conclusions.Lastly, we question the utility of layers of homology and the meaning of homology at the character state level in the context of sequence evolution.From this, we move forward to present an idealized strategy for characterizing gene family evolution for both systematic and functional purposes, including recent methodological improvements.

View Article: PubMed Central - HTML - PubMed

Affiliation: Graduate School of Biomedical Sciences, UMDNJ. liberles@uwyo.edu

ABSTRACT
Gene families are widely used in comparative genomics, molecular evolution, and in systematics. However, they are constructed in different manners, their data analyzed and interpreted differently, with different underlying assumptions, leading to sometimes divergent conclusions. In systematics, concepts like monophyly and the dichotomy between homoplasy and homology have been central to the analysis of phylogenies. We critique the traditional use of such concepts as applied to gene families and give examples of incorrect inferences they may lead to. Operational definitions that have emerged within functional genomics are contrasted with the common formal definitions derived from systematics. Lastly, we question the utility of layers of homology and the meaning of homology at the character state level in the context of sequence evolution. From this, we move forward to present an idealized strategy for characterizing gene family evolution for both systematic and functional purposes, including recent methodological improvements.

No MeSH data available.