Limits...
The tree versus the forest: the fungal tree of life and the topological diversity within the yeast phylome.

Marcet-Houben M, Gabaldón T - PLoS ONE (2009)

Bottom Line: However, such approach has been criticized for not being able to properly represent the topological diversity found among gene trees.We found that, despite high levels of among-gene topological variation, the species trees do represent widely supported phylogenetic relationships.Finally, we discuss the implications of the high levels of topological variation for phylogeny-based orthology prediction strategies.

View Article: PubMed Central - PubMed

Affiliation: Bioinformatics Department, Centro de Investigación Príncipe Felipe, València, Spain.

ABSTRACT
A recurrent topic in phylogenomics is the combination of various sequence alignments to reconstruct a tree that describes the evolutionary relationships within a group of species. However, such approach has been criticized for not being able to properly represent the topological diversity found among gene trees. To evaluate the representativeness of species trees based on concatenated alignments, we reconstruct several fungal species trees and compare them with the complete collection of phylogenies of genes encoded in the Saccharomyces cerevisiae genome. We found that, despite high levels of among-gene topological variation, the species trees do represent widely supported phylogenetic relationships. Most topological discrepancies between gene and species trees are concentrated in certain conflicting nodes. We propose to map such information on the species tree so that it accounts for the levels of congruence across the genome. We identified the lack of sufficient accuracy of current alignment and phylogenetic methods as an important source for the topological diversity encountered among gene trees. Finally, we discuss the implications of the high levels of topological variation for phylogeny-based orthology prediction strategies.

Show MeSH
Comparison of different orthology inference algorithms.The synteny based and manually curated orthology predictions available at YGOB database [18] is taken as a golden set to compute the number of true positives (TP), false positives (FP) and false negatives (FN) yielded by each method. For each method, the sensitivity S = TP/(TP+FN) and the positive predictive value P = TP/(TP+FP) are computed.
© Copyright Policy
Related In: Results  -  Collection


getmorefigures.php?uid=PMC2629814&req=5

pone-0004357-g002: Comparison of different orthology inference algorithms.The synteny based and manually curated orthology predictions available at YGOB database [18] is taken as a golden set to compute the number of true positives (TP), false positives (FP) and false negatives (FN) yielded by each method. For each method, the sensitivity S = TP/(TP+FN) and the positive predictive value P = TP/(TP+FP) are computed.

Mentions: Besides the reconstruction of species phylogenies, the existing high degree of topological variability in genome-wide data is likely to affect other applications of large-scale phylogenetic analyses. One of such applications is the large scale inference of phylogeny-based orthology predictions [12], [15], [16]. Such phylogeny-based methods are being increasingly used and are considered more accurate than standard pair-wise based methodologies [16]. There are two main approaches to infer orthology relationships from phylogenetic trees, namely reconciliation with the species tree [17] and the use of species overlap information to ascertain whether a node represents a duplication or speciation event [4]. We previously suggested that species-overlap algorithms would be more appropriate to cope with the topological diversity in single-gene phylogenies [4]. To test this, we applied both a strict tree reconciliation method and our previously described species-overlap algorithm to predict orthology relationships of all yeast genes. The orthology predictions from both methods were compared with the high-quality synteny-based orthology predictions from YGOB [18]. Although we observed no major differences in terms of positive predictive values between the two methods, there is a significant increase in terms of sensitivity when the species overlap algorithm is used (figure 2). This algorithm correctly predicted 82–96% of the true orthology relationships as compared to 32–65% values reached by species reconciliation, indicating that a relaxed consideration of tree topology is more appropriate.


The tree versus the forest: the fungal tree of life and the topological diversity within the yeast phylome.

Marcet-Houben M, Gabaldón T - PLoS ONE (2009)

Comparison of different orthology inference algorithms.The synteny based and manually curated orthology predictions available at YGOB database [18] is taken as a golden set to compute the number of true positives (TP), false positives (FP) and false negatives (FN) yielded by each method. For each method, the sensitivity S = TP/(TP+FN) and the positive predictive value P = TP/(TP+FP) are computed.
© Copyright Policy
Related In: Results  -  Collection

Show All Figures
getmorefigures.php?uid=PMC2629814&req=5

pone-0004357-g002: Comparison of different orthology inference algorithms.The synteny based and manually curated orthology predictions available at YGOB database [18] is taken as a golden set to compute the number of true positives (TP), false positives (FP) and false negatives (FN) yielded by each method. For each method, the sensitivity S = TP/(TP+FN) and the positive predictive value P = TP/(TP+FP) are computed.
Mentions: Besides the reconstruction of species phylogenies, the existing high degree of topological variability in genome-wide data is likely to affect other applications of large-scale phylogenetic analyses. One of such applications is the large scale inference of phylogeny-based orthology predictions [12], [15], [16]. Such phylogeny-based methods are being increasingly used and are considered more accurate than standard pair-wise based methodologies [16]. There are two main approaches to infer orthology relationships from phylogenetic trees, namely reconciliation with the species tree [17] and the use of species overlap information to ascertain whether a node represents a duplication or speciation event [4]. We previously suggested that species-overlap algorithms would be more appropriate to cope with the topological diversity in single-gene phylogenies [4]. To test this, we applied both a strict tree reconciliation method and our previously described species-overlap algorithm to predict orthology relationships of all yeast genes. The orthology predictions from both methods were compared with the high-quality synteny-based orthology predictions from YGOB [18]. Although we observed no major differences in terms of positive predictive values between the two methods, there is a significant increase in terms of sensitivity when the species overlap algorithm is used (figure 2). This algorithm correctly predicted 82–96% of the true orthology relationships as compared to 32–65% values reached by species reconciliation, indicating that a relaxed consideration of tree topology is more appropriate.

Bottom Line: However, such approach has been criticized for not being able to properly represent the topological diversity found among gene trees.We found that, despite high levels of among-gene topological variation, the species trees do represent widely supported phylogenetic relationships.Finally, we discuss the implications of the high levels of topological variation for phylogeny-based orthology prediction strategies.

View Article: PubMed Central - PubMed

Affiliation: Bioinformatics Department, Centro de Investigación Príncipe Felipe, València, Spain.

ABSTRACT
A recurrent topic in phylogenomics is the combination of various sequence alignments to reconstruct a tree that describes the evolutionary relationships within a group of species. However, such approach has been criticized for not being able to properly represent the topological diversity found among gene trees. To evaluate the representativeness of species trees based on concatenated alignments, we reconstruct several fungal species trees and compare them with the complete collection of phylogenies of genes encoded in the Saccharomyces cerevisiae genome. We found that, despite high levels of among-gene topological variation, the species trees do represent widely supported phylogenetic relationships. Most topological discrepancies between gene and species trees are concentrated in certain conflicting nodes. We propose to map such information on the species tree so that it accounts for the levels of congruence across the genome. We identified the lack of sufficient accuracy of current alignment and phylogenetic methods as an important source for the topological diversity encountered among gene trees. Finally, we discuss the implications of the high levels of topological variation for phylogeny-based orthology prediction strategies.

Show MeSH