• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of genoresGenome ResearchCSHL PressJournal HomeSubscriptionseTOC AlertsBioSupplyNet
Genome Res. Jun 2000; 10(6): 808–818.
PMCID: PMC310900

Whole-genome Trees Based on the Occurrence of Folds and Orthologs: Implications for Comparing Genomes on Different Levels

Abstract

We built whole-genome trees based on the presence or absence of particular molecular features, either orthologs or folds, in the genomes of a number of recently sequenced microorganisms. To put these genomic trees into perspective, we compared them to the traditional ribosomal phylogeny and also to trees based on the sequence similarity of individual orthologous proteins. We found that our genomic trees based on the overall occurrence of orthologs did not agree well with the traditional tree. This discrepancy, however, vanished when one restricted the tree to proteins involved in transcription and translation, not including problematic proteins involved in metabolism. Protein folds unite superficially unrelated sequence families and represent a most fundamental molecular unit described by genomes. We found that our genomic occurrence tree based on folds agreed fairly well with the traditional ribosomal phylogeny. Surprisingly, despite this overall agreement, certain classes of folds, particularly all-beta ones, had a somewhat different phylogenetic distribution. We also compared our occurrence trees to whole-genome clusters based on the composition of amino acids and di-nucleotides. Finally, we analyzed some technical aspects of genomic trees—e.g., comparing parsimony versus distance-based approaches and examining the effects of increasing numbers of organisms. Additional information (e.g. clickable trees) is available from http://bioinfo.mbb.yale.edu/genome/trees.

Introduction: Traditional Single-gene Phylogeny

The sequencing of whole genomes of microbial organisms allows us to reassess how we place organisms into groups and relate them to each other in phylogenetic trees. Traditionally, microorganisms have been grouped together into trees based on the sequence similarity of small subunit ribosomal RNA (SSU rRNA) (Woese et al. 1990; Woese 1987). This approach uses a single important and highly conserved gene, which has complex interactions with many other RNAs and proteins, as a basis of phylogeny. Despite its popularity, there are a number of long-standing difficulties with this approach—for example, long-branch attraction, unresolved tree differences, rate variation among sites, and mutational saturation (Lopez et al. 1999; Doolittle 1999; Lawrence 1999; Jain et al. 1999; Gogarten and Olendzenski 1999). Some researchers have even proposed that rRNA itself can be horizontally transferred between organisms (Nomura 1999; Yap 1999).

Many researchers have also tried building trees based on sequence similarity of individual protein families, such as the cytochromes, ATPases, elongation factors, aminoacyl tRNA synthases, beta-tubulins, or RNA polymerases (Makarova et al. 1999; Teichmann and Mitchison 1999a; Tumbula t al. 1999; Lake et al. 1999; Doolittle 1998; Rivera et al. 1998; Ibba et al. 1999, 1997; Edlind et al. 1996; Baldauf et al. 1996; Brown and Doolittle 1995; Andersson et al. 1998; Tomb et al. 1997; Bult et al. 1996; Lake 1994). These studies have often resulted in a wide range of implied phylogenies. The differences from the accepted ribosomal phylogeny are usually attributed to such factors as horizontal transfer or the existence of ambiguous paralogs. However, sometimes they have been used to argue that the rRNA tree is not representative of the true phylogeny. This has been particularly effective when the protein tree is based on a complex and fundamental protein such as RNA polymerase, or when the protein tree is backed up by extensive natural history evidence (Hirt et al. 1999; Edlind et al. 1996).

Whole-genome Trees and the Current Controversy

Because SSU rRNA and other individual gene families each correspond to only a tiny fraction of the genomic material in most microorganisms, focusing exclusively on them ignores the bulk of the genetic information in constructing phylogenetic trees. (In particular, the ~1.8 kb of SSU rRNA makes up less than 0.2% of most microbial genomes, which are [greater, double equals]1 Mb.) Now with the advent of completely sequenced genomes, it is possible to build trees that encompass much more of the genetic information in an organism. This has led to a profusion of new approaches toward phylogenetic estimation and a heated controversy about the structure of the fundamental tree of life, which has even been featured in the popular press (Pennisi 1998, 1999; Stevens 1999). On one extreme, some prefer traditional ribosomal trees. On the other extreme, some argue that trees are not really meaningful given the widespread evidence of horizontal transfer in microorganisms and that nets or more general graph structures should be used instead (Hilario and Gogarten 1993). In between, a third perspective maintains that most microorganisms can be arranged into meaningful trees; however, these trees might not always reflect the branching pattern suggested by rRNA.

To contribute to this debate, we build trees considering progressively more and more of the information in a genome and compare them with the traditionally proposed phylogeny. In particular, we are proposing a number of novel trees based on the occurrence of specific features, either folds or orthologs, throughout the whole genome. We call these genomic trees or whole-genome trees. Our approach toward genomic trees, which is schematized in Table Table1,1, is similar to the practice in traditional phylogenetic analysis using the presence or absence of morphologic characteristics or heritable traits such as hair or vertebrae to group organisms(Hennig 1965; Maisey 1986).

Table 1
Schematization of the Different Levels

Our first set of genomic trees is based on the occurrence of orthologous proteins. This work builds upon recent work done by other researchers clustering genomes based on the occurrence of protein families (Tekaia et al. 1999; Snel et al. 1999; Teichmann and Mitchison 1999a). In particular, Tekaia et al. (1999) developed a methodology for comparing the whole proteome content of one genome against another and using the loss or acquisition of genes to build trees. We add to this work by dividing the proteome into various classes, building trees based on these, and looking at their consistency.

Our second set of genomic trees is based on protein folds. Folds group together a number of protein families that may not share sequence similarity but do share the same essential molecular shape. As each protein fold represents a unique three-dimensional shape used by an organism, folds are ideal characteristics for building phylogenetic trees. To build our fold trees, we used a similar approach to that for the ortholog trees, in this case using the presence or absence of folds in particular genomes for tree construction. Our fold tree work is an extension of our previous investigations (Gerstein 1998a). Related work comparing genomes in terms of the occurrence of protein folds has been done by Wolf et al. (1999), who used somewhat different definitions to cluster organisms based on folds.

In a strict sense, our fold occurrence and ortholog occurrence trees are partial proteomes rather than whole-genome trees, because they are based on considering simultaneously a large, but not complete, portion of the protein-coding regions of genomes. The entire genome cannot be used, because not all of the proteins can be classified as part of orthologous groups or can be assigned to fold families. In the last part of our analysis, we look at trees that are based on information from the entire genome, using amino acid and dinucleotide composition. Although these trees represent an unbiased consideration of all the information in the genome, they condense everything to a simple composition vector, discarding much important detail. Our composition tree analysis follows up on our previous work (Gerstein, 1998b) and extensive work done by Karlin and co-workers (Karlin and Burge 1995; Karlin and Mrazek 1997; Campbell et al. 1999). Finally, it is worth pointing out that a number of additional approaches toward whole-genome trees have been advanced beyond those discussed here. In particular, Gupta (1998) used insertions and deletions along with cell-membrane structures for tree reconstruction.

Note that, whereas the occurrence and composition trees have the advantage over the single-gene trees in that they incorporate more genome information, they are not as clearly associated with an evolutionary mechanism. That is, underlying the ribosomal tree there is a specific biologic mechanism, internal to each organism, generating sequence diversity: the mutation of single base pairs, which happens at a rate roughly proportional to time. However, the way a single organism expands or contracts its repertoire of folds or orthologs cannot be explained simply in terms of individual molecular events. If one explains the acquisition of folds with horizontal transfer, the tree is no longer based on ancestral characteristics evolving internally but on the interaction with other organisms. Therefore, as opposed to true evolutionary phylogenies, whole-genome trees would then more appropriately be described as clusterings.

RESULTS AND DISCUSSION

Genomes Analyzed and Tree Techniques Used

We focus on the first eight microbial genomes to be sequenced, which include representatives from all three domains of life (Table (Table2).2). All our trees were built with the standard programs using both maximum-parsimony and distance-based methods (Felsenstein 1993, 1996; Swofford 1998). In general, we found that distance-based methods gave more reasonable results, probably because of the great divergence of the taxa, which has been remarked upon in other contexts (Swofford et al. 1996). Additional information (clickable trees, plots, etc.) is available on the Internet at http://bioinfo.mbb.yale.edu/genome/trees.

Table 2
The Eight Completely Sequenced Organisms Used in this Study

Ortholog and Fold Assignments

Orthologs were selected from the COGs (clusters of orthologous groups) database (Tatusov et al. 1997; Koonin et al. 1998). This database lists groups of ortholog sequences in eight of the first genomes sequenced based on whether or not they form a mutually consistent sequence of best matches between genomes. This database is in wide use and is accepted as a valid source of functional annotation. It groups orthologs into a hierarchy of functional classes–(e.g. class J, transcription, is part of the “Information Storage and Processing” superclass). The presence or absence of specific proteins for all the COGs for different genomes can be derived from the web site data files.

Folds were assigned to the genome sequences based on a previously described approach (Gerstein 1997, 1998b; Teichmann and Mitchison 1999b; Hegyi and Gerstein 1999). We compared the structure databank (the PDB) against the genome sequences by using both pairwise and multiple-sequence methods and standard thresholds (FASTA and PSI-BLAST, Lipman and Pearson 1985; Pearson 1996; Altschul et al. 1997). We used the SCOP classification to group the domain-level structure matches into different fold families (Murzin et al. 1995). The SCOP (structural classification of proteins) classification is assembled based on expert manual judgement, and we have augmented it with our automatically derived protein-structural alignments (Gerstein and Levitt 1998). Like the COG scheme, the SCOP classification is in wide use and accepted as a reliable classification of a protein's fold.

Single-gene Trees, a Reference Point

Traditional Ribosomal RNA Trees

As a reference point, we started our survey by constructing a traditional phylogenetic tree based on the small subunit ribosomal RNA. This established a basis of comparison for the trees generated in this study. The ribosomal tree in Figure Figure1A1A resulted in three general clusters corresponding to Archaea, Bacteria, and Eukaryota. The construction of a tree based on the large subunit ribosomal RNA in Figure Figure1B1B showed some variation in the topology of the tree.

Figure 1
Representative single-gene trees. (A) The traditional small subunit ribosomal phylogenetic tree. This is a tree of eight completely sequenced representative organisms constructed with the SSU rRNA. Trees could be constructed using data from two different ...

Single-gene Ortholog Trees

Next we examined the trees based on individual orthologous proteins chosen from the COGs database. Again this was to establish a reference for comparison and also to see the variation within single-gene trees. We focused on orthologous groups for which each of our eight organisms had only one representative, to minimize the possible effects of unrecognized paralogy (Teichmann and Mitchison 1999b). As shown in Figure Figure1,1, as has been remarked on in previous studies, the clusterings exhibit great variation depending on the protein chosen. Furthermore, many of the trees have only marginal bootstrap values if we consider 95% to be the cutoff for reasonable confidence (Efron et al. 1996). In Figure Figure1C,1C, we show a representative tree that agrees well with the traditional ribosomal phylogeny. It is based on the 30S ribosomal protein S3 (COG 92, Class J). It has relatively good bootstrap values compared to other single-gene trees that we constructed and compared to the findings of others (Teichmann and Mitchison 1999a), but not all of the values are above 95%. Figure Figure1D1D shows an example of a tree that differs significantly from the traditional phylogeny, that of triosephosphate isomerase (TIM, COG 149, Class C). Perhaps predictably, we found that trees that agreed well with the traditional ribosomal phylogeny tended to be based on proteins involved in transcription and translation, especially those with extensive RNA interactions. [This latter observation is also true for some proteins, such as the signal recognition particle (SRP) GTPase, which are not involved in transcription or translation.] In contrast, soluble enzymes, such as TIM, tended to produce trees with greater variation. These discrepancies provide evidence that trees built on sequence-similarity of individual genes would result in different phylogenies, because different genes have different mutational rates and some are horizontally transferred.

Genomic Occurrence Trees

Now that we have described the single-gene perspective as a reference, we can progress to the focus of our analysis, constructing genomic trees that consider more than the variation of individual genes. These trees were defined in terms of presence or absence of shared characteristics throughout the whole genome. Broadly, these characteristics potentially could be orthologs, homologs, or folds. We focused on orthologs and folds. We used both distance-based and parsimony methods for tree construction. For the distance-based methods, we defined the distance between two organisms with a normalized Hamming distance, which was expressed as the fraction of unshared characters divided by the total number of characters in the genomes—that is (A + B−2S)/(A + B), where A and B are the characters in the first and second genomes, respectively, and S is the number of shared characters between A and B.

We had hoped that parsimony would produce reasonable trees, because this method would automatically propose ancestral organisms that had the intermediate configurations of orthologs or folds. However, we found that, in general, distance-based methods resulted in trees closer to the traditional phylogeny. This may be because of the great divergence of the organisms studied. We give an example of the superiority of distance-based methods in Figure Figure3,3, which compares fold trees based on distances and parsimony. Also, because of the divergence of the organisms, a number of our trees may show some evidence of long-branch attraction. This arises when there are differing rates of variation among different sites in a gene, resulting in the clustering of organisms with higher rates of sequence change (Felsenstein 1996). However, we feel long-branch attraction affects the ribosomal tree as much as, if not more than, the whole-genome trees, as evident in its longer relative branch lengths and more star-shaped appearance (i.e., with less well-resolved branching). Furthermore, it is known that distance-based methods are less sensitive to long-branch attraction than parsimony, perhaps suggesting why we found the distance-based trees more reasonable.

Figure 3
Genomic trees based on the occurrence of folds. (A) Genomic tree based on the overall occurrence of folds in the genomes, generated by a distance-based method. For each of the microbial organisms the presence or absence of folds was marked with 1 or 0, ...

Genomic Tree Derived from Occurrence of Orthologs

Figure Figure22 shows the trees built in terms of the overall occurrence of ortholog families in the eight genomes. We used the COGs database to determine whether a genome had a particular orthologous group. The overall clustering based on all the orthologs in the COGs database is notably different from the traditional ribosomal tree. The Mycoplasma pneumoniae and Mycoplasma genitalium cluster is placed with the Methanococcus jannaschii, and the generally conserved grouping of Escherichia coli and Haemophilus influenzae does not occur.

Figure 2
Genomic trees based on the occurrence of orthologs. (A) Distance-based genomic tree based on the overall occurrence of orthologous proteins in the complete genome. One of the alternative methods we used for phylogenetic analysis involved building trees ...

Subdivision by Functional Class

The COGs database contains three main functional subdivisions: metabolism, information storage and processing, and cellular processes. For convenience, we will refer to these as the metabolic, information, and cellular subdivisions. More than half of the total ortholog groups and half of the signal in the overall tree come from proteins in the metabolic subset. If we remove the metabolic subset, we get a tree much more consistent with the traditional phylogeny. Figure Figure2C2C shows a genomic tree based on the occurrence of orthologs just in the information set. Its topology is almost identical to the traditional tree, the only difference being the placement of Synechocystis and Helicobacter pylori, which are reversed. This difference is minor as the divergence between these two organisms is very small even in the traditional tree. One sees, furthermore, that the traditional topology is preserved even when one selects a subset of the information set, namely the orthologs involved in transcription (class J) in Figure Figure22F.

In contrast to the information-subset tree, the metabolic-subset tree in Figure Figure2B2B corresponds closely to the topology of the overall ortholog-occurrence tree, dominating it and giving rise to its nontraditional topology. This unusual tree topology is accentuated even more when we look at a subset of the metabolic proteins corresponding to less essential functions that are less evenly maintained across organisms. This is shown in Figure Figure2E,2E, which shows a tree with 65 COGs involved in coenzyme metabolism (class H).

The topology of the metabolic subset, in fact, seems skewed by the number of orthologous proteins that each genome in this subset has. The two genomes with the largest number of orthologous proteins in the subset, E. coli (350) and Synechocystis (330), were grouped together. Haemophilus influenzae (264), having the third highest number of clusters of orthologous metabolic proteins, branches off from this cluster, followed by S. cerevisiae (262), H. pylori (226), M. jannaschii (215), and the mycoplasmas (81 and 75). We tried a variety of alternative distance measures to correct for this effect (e.g., by dividing the number of shared orthologous groups over the number of groups only present in one) but were unsuccessful in getting the traditional topology from the metabolic subset.

Thus, our results suggest that the occurrence of proteins associated with transcription and translation is closer to the traditional rRNA phylogeny than that associated with metabolism. These outcomes are reasonable and in consonance with the results that show that trees based on individual proteins involved in metabolism are usually farther from the established phylogeny than those based on proteins involved with transcription (see references above).

Fold Trees

In Figure Figure33 we show trees based on the occurrence of protein folds throughout the genome. Folds unite protein families that share the same basic architecture but which might not have any appreciable sequence similarity. They provide an ideal type of character to use in the construction of occurrence trees, since it is believed that there are only a very limited number of protein folds (Chothia 1992). The occurrence of a particular fold within the genome represents the organism selecting a particular three-dimensional shape from the overall master parts list found in nature. Fold trees have the advantage over ortholog trees in that the assignment of a particular ORF to a fold can be done fairly automatically and objectively, whereas the assignment of ORFs to various orthologous groups is often more ambiguous and requires considerable manual intervention.

Our overall fold tree is shown in Figure Figure3A.3A. It has a remarkably similar topology to the traditional ribosomal tree, especially when one considers how radically different the criteria are for building the trees. The tree shown in Figure Figure3A3A is built with distance-based methods, using a normalized Hamming distance (the number of different folds between genomes as a fraction of their total). As a contrast, in Figure Figure3B3B we show a fold tree built on the basis of parsimony. The parsimony tree clearly differs from the distance-based tree. Although these two trees were still similar, the parsimony tree alternates the position of H. pylori and S. cerevisiae, placing the latter much closer to the E. coli, H. influenzaee, Synechocystis bacterial cluster than to the Archaea, M. jannaschii.

Subdivision by Fold Class

As with the orthologous protein trees, we analyzed the composition of the total tree through various subdivisions. We can subdivide folds into four major structural classes, all-alpha, all-beta, alpha/beta, and alpha+beta (Levitt and Chothia 1976). One might not expect much variation between the classes, since unlike the orthologous proteins whose subsets had definite functional and evolutionary implications, the fold class subdivisions are not clearly related to any of these aspects. However, the trees are, in fact, different among the fold structural class subdivisions, as seen in Figure Figure33.

While the alpha+beta and all-alpha subdivisions look most similar to the overall fold tree and the traditional phylogeny, the all-beta class has an unusual clustering. Specifically, E. coli is not placed with H. influenzae, and S. cerevisiae is placed deep within the bacterial cluster. Perhaps this reflects the less even distribution of all-beta proteins and their selective proliferation in various taxa. It has been suggested before, based on bulk structure prediction, that there seems to be a different distribution of all-beta proteins in eukaryotes than prokaryotes (Gerstein 1997).

Genome Composition Trees

Because we cannot assign every single ORF in a genome to a known fold or ortholog family, the genomic occurrence trees based on these characters are in a strict sense partial proteome trees. Consequently, they suffer from potential biases because the orthologs or folds that are selected may not represent a truly random sampling of proteins in the genomes. One simple way of building a tree that takes into account all the information within the genome in unbiased fashion is using the overall nucleotide composition. To complete our analysis, we built trees based on dinucleotide and amino acid (essentially trinucleotide) composition. We constructed vectors representing the normalized composition of the genome, denoted by f, and then took the difference of these vectors to define our tree. We thus used the following relation for the distance between genomes i and j:

equation M1

where f(i,k) represents the composition and the sum over k runs from 1 to M=16 or M=20, depending on whether the tree is for dinucleotide composition or amino acid. Clearly, in the computation of composition, while we are broadly considering the entire genome, we are discarding much information by reducing the entire genome into a single composition vector.

Dinucleotide Composition

Dinucleotide composition results in a tree that is very different from the traditional tree. As seen in Figure Figure4A,4A, the clustering did not show any of the patterns observed in most of the trees in this paper. Even M. pneumoniae and M. genitalium were not clustered together.

Figure 4
Trees based on overall composition. (A) Dinucleotide composition tree. We counted the relative frequency of the dinucleotides for the complete genomes of the eight organisms. Distance between two species of dinucleotides is the distance between the 16-dimensional ...

Amino Acid Composition

The amino acid composition tree contained great similarity to other trees presented in this survey. The mycoplasmas are clustered together, as are E. coli, Synechocystis, and H. influenzae; M. jannaschii is far from the main clusters. As can be seen in Figure Figure4B,4B, simple amino acid composition, for even the entire sequence, cannot generate anything like the traditional tree; it is noteworthy that by changing from two nucleotides in the dinucleotide analysis to three in the amino acid analysis (and taking into consideration reading frames) much more information is revealed.

CONCLUSION

We built trees grouping organisms based on the overall occurrence of molecular features throughout their genomes. Most broadly, these characteristics could be orthologs, homologs, or folds. We focused on orthologs and folds. For folds we found that the overall genomic tree agreed surprisingly well with the traditional ribosomal tree. However, the distribution of all-beta folds was somewhat different. For orthologs we found that an occurrence tree based just on proteins involved in transcription and translation also agreed quite well with the traditional phylogeny. However, one built based on metabolic proteins had a rather skewed topology, and as the metabolic subset comprised most of orthologs, the overall ortholog tree also shared this appearance. This implies that adding more features does not necessarily increase the accuracy of the tree, an observation that, of course, has been made numerous times in relation to traditional phylogeny (Swofford et al. 1996). We compared our occurrence trees with many other possible trees, ranging from single-gene trees based on sequence similarity of individual orthologous proteins to entire-genome composition trees. We found that many of these alternate trees had rather unusual topologies, providing a good context for appreciating how remarkable was the agreement between whole-genome occurrence trees, particularly the fold tree, and the traditional tree.

Prospects: Trees Based on More Genomes

The number of genomes being sequenced is increasing at a fast rate and obviously the eight-genome scale of analysis presented here will be soon out of date. However, the real question is whether our approach of comparing genomes in terms of the occurrence of orthologs or folds will scale with more and more organisms. We believe it will. Recently, we have built whole-genome occurrence trees based on more than eight genomes and found that they had quite a reasonable topology. As an illustration, in Figure Figure5,5, we show a fold tree based on 20 genomes.

Figure 5
Prospects for the future. The figure shows a 20-genome tree based on the occurrence of folds. This is similar to Figure Figure3A.3A. The unit in the SCOP classification that was used was the structural superfamily rather than the fold. For eight ...

Acknowledgments

We thank H. Hegyi for help with the fold assigments and G. Naylor for discussions on phylogeny. M.G. thanks the NIH and the Donaghue Foundation for support.

The publication costs of this article were defrayed in part by payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 USC section 1734 solely to indicate this fact.

Footnotes

E-MAIL ude.elay@nietsreG.kraM; FAX (203) 432-5175.

REFERENCES

  • Altschul SF, Madden TL, Schaffer AA, Zhang J, Zhang Z, Miller W, Lipman DJ. Gapped BLAST and PSI-BLAST: A new generation of protein database search programs. Nucleic Acids Res. 1997;25:3389–3402. [PMC free article] [PubMed]
  • Andersson SG, Zomorodipour A, Andersson JO, Sicheritz-Ponten T, Alsmark UC, Podowski RM, Naslund AK, Eriksson AS, Winkler HH, Kurland CG. The genome sequence of Rickettsia prowazekii and the origin of mitochondria. Nature. 1998;396:133–140. [PubMed]
  • Baldauf SL, Palmer JD, Doolittle WF. The root of the universal tree and the origin of eukaryotes based on elongation factor phylogeny. Proc Natl Acad Sci USA. 1996;93:7749–7754. [PMC free article] [PubMed]
  • Blattner FR, Plunkett G, III, Bloch CA, Perna NT, Burland V, Riley M, Collado-Vides J, Glasner JD, Rode CK, Mayhew GF, et al. The complete genome sequence of Escherichia coli K-12. Science. 1997;277:1453–1462. [PubMed]
  • Brown JR, Doolittle WF. Root of the universal tree of life based on ancient aminoacyl-tRNA synthetase gene duplications. Proc Natl Acad Sci USA. 1995;92:2441–2445. [PMC free article] [PubMed]
  • Bult CJ, White O, Olsen GJ, Zhou L, Fleischmann RD, Sutton GG, Blake JA, FitzGerald LM, Clayton RA, Gocayne JD, et al. Complete genome sequence of the methanogenic archaeon, Methanococcus jannaschii. Science. 1996;273:1058–1073. [PubMed]
  • Campbell A, Mrazek J, Karlin S. Genome signature comparisons among prokaryote, plasmid, and mitochondrial DNA. Proc Natl Acad Sci USA. 1999;96:9184–9189. [PMC free article] [PubMed]
  • Chothia C. Proteins—1000 families for the molecular biologist. Nature. 1992;357:543–544. [PubMed]
  • De Rijk P, Robbrecht E, de Hoog S, Caers A, Van de Peer Y, De Wachter R. Database on the structure of large subunit ribosomal RNA. Nucleic Acids Res. 1999;27:174–178. [PMC free article] [PubMed]
  • Doolittle RF. Microbial genomes opened up. Nature. 1998;392:339–342. [PubMed]
  • Doolittle WF. Phylogenetic classification and the universal tree [see comments] Science. 1999;284:2124–2129. [PubMed]
  • Edlind TD, Li J, Visvesvara GS, Vodkin MH, McLaughlin GL, Katiyar SK. Phylogenetic analysis of beta-tubulin sequences from amitochondrial protozoa. Mol Phylogenet Evol. 1996;5:359–367. [PubMed]
  • Efron B, Halloran E, Holmes S. Bootstrap confidence levels for phylogenetic trees. Proc Natl Acad Sci USA. 1996;96:13,429–13,434.
  • Felsenstein, J. 1993. PHYLIP (Phylogeny Inference Package) version 3.5c. Distributed by the author. Department of Genetics, University of Washington, Seattle.
  • ————— Inferring phylogenies from protein sequences by parsimony, distance, and likelihood methods. Meth Enzymol. 1996;266:418–427. [PubMed]
  • Fleischmann RD, Adams MD, White O, Clayton RA, Kirkness EF, Kerlavage AR, Bult CJ, Tomb JF, Dougherty BA, Merrick J, et al. Whole-genome random sequencing and assembly of haemophilus influenzae rd. Science. 1995;269:496–512. [PubMed]
  • Fraser CM, Gocayne JD, White O, Adams MD, Clayton RA, Fleischmann RD, Bult CJ, Kerlavage AR, Sutton G, Kelley JM, et al. The minimal gene complement of Mycoplasma genitalium. Science. 1995;270:397–403. [PubMed]
  • Gerstein M, Levitt M. Comprehensive assessment of automatic structural alignment against a manual standard, the Scop classification of proteins. Protein Sci. 1998;7:445–456. [PMC free article] [PubMed]
  • Gerstein M. A structural census of genomes: Comparing eukaryotic, bacterial and Archaeal genomes in terms of protein structure. J Mol Biol. 1997;274:562–576. [PubMed]
  • Gerstein M. How representative are the known structures of the proteins in a complete genome? A comprehensive structural census. Fold Des. 1998a;3:497–512. [PubMed]
  • Gerstein M. Patterns of protein-fold usage in eight microbial genomes: A comprehensive structural census. Proteins. 1998b;33:518–534. [PubMed]
  • Goffeau A, Aert R, Agostini-Carbone ML, Ahmed A, et al. The yeast genome directory. Nature. 1997;387(Supp):5–105.
  • Gogarten JP, Olendzenski L. Orthologs, paralogs and genome comparisons. Curr Opin Genet Dev. 1999;9:630–636. [PubMed]
  • Gupta RS. Protein phylogenies and signature sequences: A reappraisal of evolutionary relationships among archaebacteria, eubacteria, and eukaryotes. Microbiol Mol Biol Rev. 1998;62:1435–1491. [PMC free article] [PubMed]
  • Hegyi H, Gerstein M. The relationship between protein structure and function: A comprehensive survey with application to the yeast genome. J Mol Biol. 1999;288:147–164. [PubMed]
  • Hennig W. Phylogenetic systematics. Ann Rev Entomol. 1965;10:97–116.
  • Hilario E, Gogarten JP. Horizontal transfer of ATPase genes—the tree of life becomes a net of life. Biosystems. 1993;31:111–119. [PubMed]
  • Himmelreich R, Hilbert H, Plagens H, Pirkl E, Li BC, Herrmann R. Complete sequence analysis of the genome of the bacterium Mycoplasma pneumoniae. Nucleic Acids Res. 1996;24:4420–4449. [PMC free article] [PubMed]
  • Hirt RP, Logsdon JM, Jr, Healy B, Dorey MW, Doolittle WF, Embley TM. Microsporidia are related to fungi: Evidence from the largest subunit of RNA polymerase II and other proteins. Proc Natl Acad Sci USA. 1999;96:580–585. [PMC free article] [PubMed]
  • Ibba M, Losey HC, Kawarabayasi Y, Kikuchi H, Bunjun S, Soll D. Substrate recognition by class I lysyl-tRNA synthetases: A molecular basis for gene displacement. Proc Natl Acad Sci USA. 1999;96:418–423. [PMC free article] [PubMed]
  • Ibba M, Morgan S, Curnow AW, Pridmore DR, Vothknecht UC, Gardner W, Lin W, Woese CR, Soll D. A euryarchaeal lysyl-tRNA synthetase: Resemblance to class I synthetases. Science. 1997;278:1119–1122. [PubMed]
  • Jain R, Rivera MC, Lake JA. Horizontal gene transfer among genomes: The complexity hypothesis. Proc Natl Acad Sci USA. 1999;96:3801–3806. [PMC free article] [PubMed]
  • Kaneko T, Sato S, Kotani H, Tanaka A, Asamizu E, Nakamura Y, Miyajima N, Hirosawa M, Sugiura M, Sasamoto S, et al. Sequence analysis of the genome of the unicellular cyanobacterium Synechocystis sp. strain PCC6803. II. Sequence determination of the entire genome and assignment of potential protein-coding regions (supplement) DNA Res. 1996;3:185–209. [PubMed]
  • Karlin S, Burge C. Dinucleotide relative abundance extremes: A genomic signature. [Review] Trends Genetics. 1995;11:283–290. [PubMed]
  • Karlin S, Mrazek J. Compositional differences within and between eukaryotic genomes. Proc Natl Acad Sci USA. 1997;94:10,227–10,232. [PMC free article] [PubMed]
  • Koonin EV, Tatusov RL, Galperin MY. Beyond complete genomes: From sequence to structure and function. Curr Opin Struct Biol. 1998;8:355–363. [PubMed]
  • Lake JA. Reconstructing evolutionary trees from DNA and protein sequences: Paralinear distances. Proc Natl Acad Sci USA. 1994;91:1455–1459. [PMC free article] [PubMed]
  • Lake JA, Jain R, Rivera MC. Mix and match in the tree of life. Science. 1999;283:2027–2028. [PubMed]
  • Lawrence JG. Gene transfer, speciation, and the evolution of bacterial genomes. Curr Opin Microbiol. 1999;2:519–523. [PubMed]
  • Levitt M, Chothia C. Structural patterns in globular proteins. Nature. 1976;261:552–558. [PubMed]
  • Lipman DJ, Pearson WR. Rapid and sensitive protein similarity searches. Science. 1985;227:1435–1441. [PubMed]
  • Lopez P, Forterre P, Philippe H. The root of the tree of life in the light of the covarion model. J Mol Evol. 1999;49:496–508. [PubMed]
  • Maidak BL, Cole JR, Parker CT, Jr, Garrity GM, Larsen N, Li B, Lilburn TG, McCaughey MJ, Olsen GJ, Overbeek, R. R, et al. A new version of the RDP (Ribosomal Database Project) Nucleic Acids Res. 1999;27:171–173. [PMC free article] [PubMed]
  • Maisey JG. Heads and tails: A chordate phylogeny. Cladistics. 1986;2:201–256.
  • Makarova KS, Aravind L, Galperin MY, Grishin NV, Tatusov RL, Wolf YI, Koonin EV. Comparative genomics of the Archaea (Euryarchaeota): Evolution of conserved protein families, the stable core, and the variable shell. Genome Res. 1999;9:608–628. [PubMed]
  • Murzin A, Brenner SE, Hubbard T, Chothia C. SCOP: A structural classification of proteins for the investigation of sequences and structures. J Mol Biol. 1995;247:536–540. [PubMed]
  • Nomura M. Engineering of bacterial ribosomes: Replacement of all seven Escherichia coli rRNA operons by a single plasmid-encoded operon [see comments] Proc Natl Acad Sci USA. 1999;96:1820–1822. [PMC free article] [PubMed]
  • Page R D. TreeView: An application to display phylogenetic trees on personal computers. Comput Appl Biosci. 1996;12:357–358. [PubMed]
  • Pearson WR. Using the FASTA program to search protein and DNA sequence databases. Meth Mol Biol. 1994;24:307–331. [PubMed]
  • ————— Effective protein sequence comparison. Meth Enzymol. 1996;266:227–258. [PubMed]
  • Pennisi E. Genome data shake tree of life [news] Science. 1998;280:672–674. [PubMed]
  • ————— Is it time to uproot the tree of life? [news] Science. 1999;284:1305–1307. [PubMed]
  • Rivera MC, Jain R, Moore JE, Lake JA. Genomic evidence for two functionally distinct gene classes. Genetics. 1998;11:6239–6244. [PMC free article] [PubMed]
  • Snel B, Bork P, Huynen MA. Genome phylogeny based on gene content. Nat Genet. 1999;21:108–110. [PubMed]
  • Stevens, W.K. 1999. Rearranging the branches on a new tree of life. The New York Times: Aug. 31 1999: F1.
  • Swofford DL. PAUP*. Phylogenetic Analysis Using Parsimony (*;and Other Methods). Version 4. Sunderland, MA: Sinauer Associates; 1998.
  • Swofford D, Olsen GJ, Waddell PJ, Hillis DM. Molecular Systematics. Sunderland, MA: Sinauer Associates; 1996. Phylogenetic Inference; pp. 407–514.
  • Tatusov RL, Koonin EV, Lipman DJ. A genomic perspective on protein families. Science. 1997;278:631–637. [PubMed]
  • Teichmann SA, Mitchison G. Is there a phylogenetic signal in prokaryote proteins? J Mol Evol. 1999a;49:98–107. [PubMed]
  • ————— Making family trees from gene families. Nat Genet. 1999b;21:66–67. [PubMed]
  • Tekaia F, Lazcano A, Dujon B. The genomic tree as revealed from whole proteome comparisons. Genome Res. 1999;9:550–557. [PMC free article] [PubMed]
  • Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F, Higgins DG. The CLUSTAL_X windows interface: Flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 1997;25:4876–4882. [PMC free article] [PubMed]
  • Tomb J-F, White O, Kerlavage AR, Clayton RA, Sutton GG, Fleischmann RD, Ketchum KA, Klenk HP, Gill S, Dougherty BA, et al. The complete genome sequence of the gastric pathogen Helicobacter pylori. Nature. 1997;388:539–547. [PubMed]
  • Tumbula D, Vothknecht UC, Kim H, Ibba M, Min B, Li T, Pelaschier J, Stathopoulos C, Becker H, Soll D. Archaeal aminoacyl-tRNA synthesis. Diversity replaces dogma [In Process Citation] Genetics. 1999;152:1269–1276. [PMC free article] [PubMed]
  • Van de Peer Y, Robbrecht E, de Hoog S, Caers A, De Rijk P, De Wachter R. Database on the structure of small subunit ribosomal RNA. Nucleic Acids Res. 1999;27:179–183. [PMC free article] [PubMed]
  • Woese CR. Bacterial evolution. Microbiol Rev. 1987;51:221–271. [PMC free article] [PubMed]
  • Woese CR, Kandler O, Wheelis ML. Towards a natural system of organisms: Proposal for the domains Archaea, Bacteria, and Eucarya. Proc Natl Acad Sci USA. 1990;87:4576–4579. [PMC free article] [PubMed]
  • Wolf YI, Brenner SE, Bash PA, Koonin EV. Distribution of protein folds in the three superkingdoms of life. Genome Res. 1999;9:17–26. [PubMed]
  • Yap WH, Zhang Z, Wang Y. Distinct types of rRNA operons exist in the genome of the actinomycete Thermomonospora chromogena and evidence for horizontal transfer of an entire rRNA operon. J Bacteriol. 1999;181:5201–5209. [PMC free article] [PubMed]

Articles from Genome Research are provided here courtesy of Cold Spring Harbor Laboratory Press
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...