CVTree3 Web Server for Whole-genome-based and Alignment-free Prokaryotic Phylogeny and Taxonomy

Gunghong ZuoBilin Ho*b

T-Life Research Center,Department of Physics,Fudan University,Shanghai 200433,China

APPLICATION NOTE

CVTree3 Web Server for Whole-genome-based and Alignment-free Prokaryotic Phylogeny and Taxonomy

Guanghong Zuoa,Bailin Hao*,b

T-Life Research Center,Department of Physics,Fudan University,Shanghai 200433,China

Available online 10 November 2015

Handled by Jingfa Xiao

Composition vector; CVTree; Whole-genome-based tree; Alignment-free phylogeny; Archaea and Bacteria taxonomy

A faithfulphylogeny and an objective taxonomy for prokaryotes should agree with each other and ultimately follow the genome data.With the number of sequenced genomes reaching tens of thousands,both tree inference and detailed comparison with taxonomy are great challenges.We now provide one solution in the latest Release 3.0 of the alignment-free and whole-genome-based web server CVTree3.The server resides in a cluster of64 cores and is equipped with an interactive, collapsible,and expandable tree display.It is capable of comparing the tree branching order with prokaryotic classification at all taxonomic ranks from domains down to species and strains. CVTree3 allows for inquiry by taxon names and trial on lineage modifications.In addition,it reports a summary of monophyletic and non-monophyletic taxa at all ranks as well as produces print-quality subtree figures.After giving an overview of retrospective verification of the CVTree approach,the power of the new server is described for the mega-classification of prokaryotes and determination oftaxonomic placement ofsome newly-sequenced genomes.Afew discrepancies between CVTree and 16S rRNA analyses are also summarized with regard to possible taxonomic revisions.CVTree3 is freely accessible to all users at http://tlife.fudan.edu.cn/cvtree3/without login requirements.

Introduction

Prokaryotes are the most abundant and successful organisms on Earth[1].However,their phylogeny and taxonomic classification had been a long-standing challenge until Carl Woese and coworkers suggested using the small subunit(SSU or 16S)rRNA sequences as molecular markers in the late 1970s [2].The completion of the second edition of the Bergey’s Manual of Systematic Bacteriology[3](hereafter referred toas the Manual)marked a culmination of16S rRNA analysis as the Manual‘‘follows a phylogenetic framework based on analysis of the nucleotide sequence of the small ribosomal subunit RNA,rather than a phenotype structure”(George Garrity’s preface).As early as in 1985,Woese et al[4] proposed a phylogenetic definition for the major eubacterial taxa using all available 16S rRNA sequences,about 400 in total.

The challenge in‘‘congruence”of prokaryotic phylogeny and taxonomy on the basis of SSU rRNA analysis,however, raises a question of principle.In order to establish an objective and valid classification of microbes,the present 16S rRNA-based scheme needs cross-verification.By all means the verification should follow the genomic data.In fact,an Ad Hoc Committee on Reconciliation of Approaches to Bacterial Systematics stated in a 1987 Report that‘‘there was general agreement that the complete DNA sequence would be the reference standard to determine phylogeny and that phylogeny should determine taxonomy”[5].Since then, genome-based phylogeny and taxonomy studies have been touched on by many research groups from different angles [6–10].

From a taxonomic perspective,although only a small fraction of the genomes sequenced so far corresponds to prokaryotes with a type strain(1725 among 12,000[10]as of early 2014),an ambitious program to sequence a myriad of type strains,known as the Genomic Encyclopedia of Bacteria and Archaea project(GEBA)[11,12],has made rapid progress. The taxonomic coverage of sequenced genomes willsoon catch up with that of the 16S rRNA collection.

Only a quarter of a century since the release of Ad Hoc Committee report,with the total number of sequenced microbialgenomes reaching tens of thousands,the materialization of‘‘general agreement”has become feasible.However, effectively inferring phylogenetic trees from the genome sequences and comparing the branching orders with taxonomy at all ranks present challenges.To this end,the composition vector(CV)approach to prokaryotic phylogeny developed by our group in the last decade[13–24]has the potential to meet this challenge.In this paper,we describe the latestversion (Version 3)of the CVTree web server and demonstrate its applications.Detailed application of CVTree3 to various aspects of microbiology will be presented in subsequent publications.

Methods

Since the methodology and foundation of the CVTree approach has been demonstrated in numerous previous publications[13–24]as well as discussed by other authors [25,26],we provide only a brief summary of the essentials in this paper.

CVTree approach

CVTree uses the whole genomes as input,thus avoiding ambiguities in selecting orthologous genes and circumventing the problem of lateral gene transfers.Whole-genome comparison must be alignment-free,as prokaryotic genomes differ significantly in their sizes and gene contents.Our method for conducting alignment-free comparison consists of extending the amino acid alphabet counts to counting the number of K-peptides in all protein products encoded in a genome.In order to highlight the shaping role of natural selection,the original counts are modified by subtracting the random background caused by neutral mutations using a(K–2)-th order Markovian prediction.

In addition to the advantages of using alignment-free data and performing whole-genome-based analyses,several distinctive features of the CVTree are listed below.

1.The peptide length K looks like a parameter but does not function as a parameter because K-values are not adjusted and the same set of K is used for all genomes in a tree.In the older versions of the CVTree server[27,28],a K-value must be set for each run.CVTree3 carries out calculations for a range of K,such as K=3–8,in a single run.Watching the branching orders with varying K provides an additionalangle for evaluating the quality of the resulting trees. We note that the best(in the sense of agreement with taxonomy)K-values are 4–5 for viruses,5–6 for prokaryotes, and 6–7 for fungi.Proof of this statement as well as a description of the role of K had been provided in previous reports[16,23].

2.Traditionally,an inferred phylogenetic tree is subject to statistical re-sampling tests such as bootstrap and jackknife analyses.However,successfully passing these tests only indicates the stability and self-consistency of the tree with respect to small variations of the input data, but not the objective correctness of the phylogeny. Though the CVTree results indeed have passed these time-consuming tests[19],we advocate the viewpoint that phylogenetic trees should be checked directly with taxonomy.

3.The comparison with taxonomy requires a reference classification scheme.In the CVTree server,each built-in genome is associated with initial lineage information taken from the NCBI Taxonomy(www.ncbi.nlm.nih.gov/taxonomy).The information is written in one line with the abbreviations〈D〉,〈K〉,〈P〉,〈C〉,〈O〉,〈F〉,〈G〉,〈S〉, and〈T〉,which stand for Domain,Kingdom,Phylum, Class,Order,Family,Genus,Species,and sTrain,respectively.A standard notation for a lacking classifier is‘‘Unclassified”.For example,‘‘〈F〉Unclassified”denotes a missing family assignment.Lineage information containing one or more‘‘Unclassified”terms is considered incomplete.

4.A central notion in comparing tree branching orders with taxonomy is monophyly.For prokaryotes,while the notion of species is still under debate,one cannot use the original definition of monophyly as the collection of descendants from one and the same common ancestor as discussed by James Farris[29,30].Moreover,monophyly is a reciprocal notion with respect to both phylogeny and taxonomy.We adopt a pragmatic approach by restricting ourselves to the input dataset and reference classification.If all genomes from one and the same taxon are represented exclusively by leaves in a single branch, the branch is said to be monophyletic.If a taxon does not appear to be monophyletic in taxonomy,e.g.,the genus Clostridium consisting of a sensu stricto cluster and several‘‘monophyletic”groups as described in volume 3 of the Manual,the corresponding branches cannot becharacterized as monophyletic.We use the term‘‘convergence”to describe these cases.A branch may converge to a monophyletic branch such as Cyanobacteria{77}or converge to several partially monophyletic clusters such as Clostridium{32/49},Clostridium{7/49},and Clostridium {4/49},meaning that the 49 genomes listed under the genus Clostridium in the reference taxonomy appear as several clusters in CVTree.

5.There are two elements of a phylogenetic tree:the branching order(topology)and the branch lengths.The former reflects taxonomy and the latter is associated with evolution.Calibration of branch lengths is always based on the assumption that the mutation rate has remained constant over evolutionary history,an assumption that cannot hold true when dealing with many phyla in a large-scale study.Although a relationship between the dis-similarity measure used in constructing the CVTree and the usual genetic distance has been derived[16],it does not always preserve the topology of the tree.Therefore,we do not scale branches in all CVTrees and only examine the branching orders.To measure evolutionary time for a group of not-too-distantly-related species,traditional methods such as multi-alignment of orthologous proteins would do the job.

CVTree3 web server

We have made the CVTree web server publicly available,so bench-biologists can take advantage of the whole-genomebased and alignment-free method.The server has been released twice:Version 1 in 2004[27]and Version 2 in 2009[28].As the CV algorithm is CPU-and memorydemanding,previous servers could not cope with the evergrowing amount of genomic data,and thus we redesigned the CVTree web server.The new CVTree3 server contains many enhanced features and is freely accessible at http:// tlife.fudan.edu.cn/cvtree3/.The main improvements are listed as follows.(1)The CV algorithm has been parallelized and the new CVTree3 pipeline now resides in a cluster with 64 cores.(2)The CVTree3 web server is not only designed as a phylogenetic tool,but also enables combined study of phylogeny and taxonomy both on a large scale across many phyla and at lower ranks down to infrasubspecific strains. (3)The server is equipped with an interactive tree display, allowing for the collapse and expansion of branches in accordance with lineage information associated with the input genomes.(4)The server reports the number of genomes in all monophyletic and non-monophyletic taxa/ branches at all ranks from the domain down to the species. (5)The server allows for trial lineage modifications and recollapsing of the tree with a new report on monophyly. (6)The server allows print-quality output of any selected subtree.

Since there is a detailed online(and printable)User’s Manual(File S1)for the web server,we will not describe in detail the aforementioned technical points except to demonstrate some useful features regarding the taxonomic placement of a few newly-sequenced genomes without proper lineage information given at the present time.

Genome resources

Inherited from the previous releases,the CVTree3 web server has a built-in genome dataset.However,because the NCBI FTP site(ftp://ftp.ncbi.nih.gov/genomes/Bacteria/)has nearly stopped releasing new bacterial genomes since the beginning of 2014,CVTree3 has to give up monthly automatic updating from the NCBI.Currently,prokaryotic genomes are collected from the NCBI,European Nucleotide Archive(ENA)at the EBI,Integrated Microbial Genomes(IMG)by the Joint Genome Institute(JGI)of the U.S.Department of Energy(DOE), the Broad Institute,the J.Craig Venter Institute,and the Pathosystems Resource Integration Center(PATRIC)(the URLs of these institutions are listed in CVTree3 User’s Manual,see File S1).In order to demonstrate the capability of the CVTree,we also included some genomes from the Microbial Dark Matter Project[31].

In this article,we refer to a fully-fledged Working Project to demonstrate the features of the CVTree3 web server.The project has a specific project number 30150127_1559_28802 in order to avoid being deleted by the web server in due time. In the input dataset,there are 342 Archaea,2870 Bacteria, and 8 Eukarya genomes.The latter includes 4 fungal and 4 non-fungal genomes serving as candidates of the out-group in tree construction.

Users may upload their own genomes together with lineage information.In the Working Project,a total of 21 genomes were uploaded and their names appeared in the‘‘Upload User Genomes and Lineage File”page.The user-supplied lineage information file carries a fixed name‘‘Lineage.txt”and it does not appear explicitly on the page.

The CVTree approach depends on genome annotation,but is insensitive to the annotation mainly because of the alignment-free methodology.For example,genomes of Gluconacetobacter diazotrophicus PAI 5 from the same source (ATCC 49037)were sequenced,assembled,and annotated by two institutions.These assemblies contained‘‘a(chǎn) surprisingly high number ofdifferences”[32],yet they appeared in CVTrees as two closely-related sisters.In addition,two genomes listed under an unclassified bacterial phylum Acetothermia from IMG JGI provided another example of the insensitivity of CVTree to annotation.Both genomes are in the permanent draft status.GenBank files may be generated from contigs using the IMG pipeline.Their names appeared in the User Uploaded Genomes in our Working Project(Acetothermia bacterium SCGC AAA255 C06 SAK 001 122 and Candidatus Acetothermum autotrophicum).By using the‘‘Search Query”function of the interactive display,the genomes appear together in CVTree at the phylum level.The fact that CVTree can accept some permanent-draft genomes greatly widens the reach of CVTree,as there are more than 23,000 permanent drafts according to the Genomes Online Database(GOLD, gold.jgi-psf.org)statistics,and this number is increasing rapidly.

Applications of CVTree3

Given an input genome set and a parameter-free method such as CVTree,an inferred tree is a fixed unchangeable subject,and the tree cannot be adjusted or modified.In contrast,taxonomy has always been a work in progress.Lineage modifications and taxonomic revisions are routine issues,leading to a convergent phylogeny-based classification of microbial organisms.Over the years,CVTree has been applied to viruses [33,34],Archaea and Bacteria[13–18,20–22,24],chloroplasts [35],and fungi[36]with remarkable success.The powerful and parallelized CVTree3 web server will bring about many additional new applications.

Upon entering the Working Project,a maximallycollapsed CVTree with three branches,corresponding to the three main domains of life,appears as shown in Figure 1. All 3220 genomes are represented in this single screen.Bacteria{2733+137}indicates that there are 137 bacterial genomes without proper lineage information.A complete lineage may also require modification in order to reflect the actual taxonomic position.By introducing lineage modifications,these numbers may change,but their sum remains at 2870.By expanding the nodes or making enquiry for a designated taxon name,any part of the tree may be unfolded for in-depth inspection.

Retrospective verifications of CVTree

Before describing the applications of CVTree3,we recall the significant fact that for prokaryotes with sequenced genomes,all taxonomic revisions or new proposals published thus far agree with CVTree or at least do not contradict the CVTree branching order.These should be regarded as retrospective verifications of the new approach.A partial list follows.

1.The move of the genus Oceanobacillus from the phylum Proteobacteria[38]to phylum Firmicutes in 2003 [39].

2.The move of the species Thiomicrospira denitrificans from the class Gammaproteobacteria to the class Epsilonproteobacteria as Sulfurimonas denitrificans in 2004[40], with the reclassification proposal published in 2006[41].

3.The reassignment of Thermoanaerobacter tengcongensis to a new genus as Caldanaerobacter tengcongensis in 2004[42].

4.The transfer of Thermomicrobium roseum from its original phylum Thermomicrobia to class Thermomicrobia in the phylum Chloroflexi in 2004[43].

5.The reclassification of Sphaerobacter thermophilus from the phylum Actinobacteria to the class Thermomicrobia in phylum Chloroflexi in 2004[43].

6.The transfer of Enterobacter sakazakii to a newlyproposed genus as Cronobacter sakazakii in 2008[44] led to a monophyletic Cronobacter{6}in the current CVTree.

Figure 1 The most collapsed CVTree with three main domains of lifeNote that all3219 genomes are visible in this single screen(a eukaryotic genome used as the outgroup was hidden).The{n+m}notation indicates that there are n genomes with complete lineage information and m genomes with incomplete or missing lineage information. {n+m}is indicated as{n}when m=0,while when n=0,{n+m}is indicated as{0+m}.〈D〉and〈K〉represent domain and kingdom,respectively.Main domains of life were defined as suggested by Woese and Fox[37].‘‘Unclassified”indicates missing classifier. Lineage information containing one or more‘‘Unclassified”is considered incomplete.

7.The reclassification ofa few Clostridium and Ruminococcus species to a newly-proposed genus Blautia in 2008 [45]led to a monophyletic Blautia{6}in current CVTree.

8.The suggestion to exclude Actinobacillus succinogenes and‘Mannheimia succiniciproducens’from their respective genera in 2008[46]and the proposal to establish a new genus Basfia to accommodate similar succinic acid-producing bacteria in 2010[47]led to three monophyletic genera Mannheimia{8},Actinobacillus{4},and Basfia{2}in CVTrees.

9.The class Mollicutes never joined the other two classes Bacilli and Clostridia of the phylum Firmicutes since the first CVTree was published in 2004.It was removed from volume 3(2009)of the Manual on Firmicutes to become a new phylum Tenericutes in volume 4 of the Manual in 2010.

10.The reclassification of Bacillus tusciae to a new genus as Kyrpidia tusciae in the family Alicyclobacillus in 2011 [48].

11.The assignment of Thermobaculum terrenum to the phylum Chloroflexi in 2011[49].

12.The move of Clostridium difficile and other clostridial species to new genera such as Peptoclostridium,Lachnoclostridium,and Ruminoclostridium in 2013[50],leading to a monophyletic Peptoclostridium{12}in the current CVTree.We note that these names are effectively published but have not been validly published,based on their absence in LPSN[51].

13.The reclassification of Agromonas oligotrophica into Bradyrhizobium oligotrophicum in 2013[52]led to a monophyletic Bradyrhizobium{6}in the current CVTree.

14.The reclassification of Thermoproteus neutrophilus to Pyrobaculum neutrophilum in 2013[53]led to two monophyletic genera Thermoproteus{2}and Pyrobaculum{8} in CVTree.

15.A recent proposal to elevate four families in the class Actinobacteria to corresponding single-family orders [54]does not contradict the current CVTree.In particular,the accommodation of the three genera Geodermatophilus,Blastococcus,and Modestobacter in the order Geodermatophilales is supported by CVTree.

16.A recent proposal to split the euryarchaeal order Halobacteria into three orders[55]is supported by CVTree[24].

To describe the possible applications of CVTree3 to microbiology,we chose a few topics to demonstrate this potential rather than to explore biologicaldetails.These include the following:large-scale classification,taxonomic placement of newly-sequenced genomes,and high resolution of CVTree at the rank species and below.

Mega-classification of prokaryotes

Large-scale classification,or as Cavalier-Smith puts it[56], mega-classification,of prokaryotes,deals with higher taxonomic ranks such as phylum,class,and order(at present, ranks higher than order are not covered by the International Bacterial Code[57]).The second edition of Bergey’s Manual [3]lists 2 archaeal and 26 bacterial phyla.The total number of prokaryotic phyla may be in the hundreds.In fact,some newly-sequenced genomes represent yet unclassified phyla or classes.The Working Project accompanying this paper helps to comprehend the overall situation.Since a comparison of the archaeal phyla has recently published[24],we concentrate on bacterial taxa.

Ifthe‘‘Modified Lineage”box in the setting-parameter page is checked,a default Lineage Modification file is used to report the convergence of taxa.If unchecked,the initial information from the NCBI Taxonomy is used.In this‘‘bare”situation,an overwhelming majority of phyla appear to be well-defined, i.e.,monophyletic for at least one K-value and occupying a position at the phylum level.These phyla include Acidobacteria {7+2},Actinobacteria{365+2},Aquificae{14},Chlamydiae {141},Chlorobi{13},Cyanobacteria{2+75},Deferribacteres {4},Deinococcus-Thermus{20},Dictyoglomi{2},Fibrobacteres {3},Fusobacteria{8},Planctomycetes{7},Synergistetes{5}, Thermotogae{19},and Verrucomicrobia{2+2}.In particular, we point at the relatively-unresolved Cyanobacteria phylogeny {2+75}.The large number(75)ofincomplete lineage information reveals the long-due challenge of Cyanobacteria taxonomy. Historically,classification of Cyanobacteria followed the Botanic Code.Currently,NCBI Taxonomy and the Bergey’s Manual differ significantly for Cyanobacteria.The complete solution ofthis issue should be examined further.

Some‘‘big”phyla,i.e.,those represented by a large number of genomes,naturally appeared to be non-monophyletic when no lineage modification was made.These include Bacteroidetes {88/89+5},Firmicutes{662},Proteobacteria{1175},and Spirochaetes{59}.We also encounter a few interesting cases,in which the CVTree results differ from 16S rRNA-based taxonomy.For example,until recently,the species Thermodesulfovibrio yellowstonii had been considered a member of the phylum Nitrospirae[58].However,in CVTree,it is resolved to the phylum Thermodesulfobacteria.This lineage modification according to CVTree would lead to monophyletic resolving of Nitrospirae{4}and Thermodesulfobacteria{3}.Similarly, the family Rhodothermaceae with its two subordinate genera Salinibacter and Rhodothermus belongs to an uncertain order (Bacteriales Order II Incertae Sedis in the Manual)of the phylum Bacteroidetes.But in CVTree,it is the nearest neighbor to the phylum Chlorobi.Corresponding lineage modification leads to monophyletic Bacteroidetes{88+5}and Chlorobi {13+5}.

The clearest distinction of the CVTree phylogeny and 16S rRNA analyses arises from the phylum Spirochaetes.Species in this phylum were placed together mainly based on their morphological similarities in the first edition of the Manual. Carl Woese described that spirochaetes form a single clade according to 16S rRNA features[4].Taxonomically,the phylum Spirochaetes consists of a single class,which in turn is composed of one order.Therefore,only the rank family makes sense.In CVTree,the three monophyletic families Spirochaetaceae{44},Brachyspiraceae{7},and Leptospiraceae{8}do not join each other,but the first two are closer together.

Three out of five classes in the phylum Proteobacteria {1175}appear to be monophyletic clusters,including Alphaproteobacteria{257+9},Betaproteobacteria{150+14}, and Epsilonproteobacteria{104+3}.Taking into account that the Beta-and Gamma-groups together form a greatermonophyletic cluster,only the class Deltaproteobacteria{61} challenges the present classification.There is a core〈C〉Deltaproteobacteria{43/61},an order〈O〉Myxococcales {13}in the neighborhood of〈P〉A(chǔ)cidobacteria,an order〈O〉Bdellovibrionales{4}joining〈F〉Leptospiraceae as a sister group,and an outlier〈G〉Hippea escaping to〈P〉A(chǔ)quificae.In a sense,only the phylum Firmicutes{662}awaits essential taxonomic revision.Historically,many phyla have been taken out from Firmicutes,e.g.,Actinobacteria and Tenericutes.The taxonomy of some genera such as Clostridium remains unsettled, although Clostridium has been modified after separating five genera in 1994[59]and six genera in 2013[50].CVTree3 may contribute to the further resolution of this problem.

（四）第三人干擾婚姻關(guān)系的行為侵犯了無(wú)過錯(cuò)配偶的配偶權(quán)。該學(xué)說認(rèn)為，第三人干擾婚姻關(guān)系的行為侵害了無(wú)過錯(cuò)方的配偶權(quán)。該學(xué)說所謂的配偶權(quán)，是指夫妻雙方在婚姻關(guān)系的存續(xù)期間、夫妻雙方所享有的作為權(quán)利和負(fù)有的不作為義務(wù)。配偶權(quán)包括了住所決定權(quán)、同居權(quán)、日常家事代理權(quán)、計(jì)劃生育義務(wù)、忠誠(chéng)義務(wù)等相對(duì)義務(wù)。配偶權(quán)相對(duì)于夫妻雙方來講是相對(duì)權(quán)，但對(duì)于婚姻關(guān)系以外的任何人也就是本文所講的“第三人”來講是絕對(duì)權(quán)。第三人不能干擾夫妻雙方所享有的性權(quán)利和同居權(quán)利，否則即構(gòu)成侵權(quán)。

Taxonomic placement of newly-sequenced genomes

The number of unclassified phyla far exceeds that of known phyla.On the SILVA web page for candidate taxonomic units (ftp.arb-silva.de/release_108/),424‘‘phyla”are numbered among one of the 15 groups in the OD1 group only.Fortunately,owing to the advent of relatively inexpensive and effective sequencing technology,these phyla are beginning to be discovered.Currently,genome-based phylogeny provides the only means for judging the taxonomic placement of genomes without phenotyping data.In CVTree,the collapsing mechanism helps group together closely-related genomes at the phylum level and above.Although not an exhaustive list,we indicate that(1)Caldiserica and Coprothermobacter(listed under Firmicutes in the Manual but considered as an‘‘established phylum”in a 2004 microbial census[60])are within the branch of〈P〉Thermotogae{19}and〈P〉Dictyoglomi{2}; and(2)Candidate division WWE3,Candidatus Saccharibacteria,and Candidatus Saccharimonas are located next to〈P〉Tenericutes{99+2}.Further expansion of the last taxon reveals{+2}to be a member of the candidate division SR1 and a misclassified delta-proteobacterium BABL1.Their relationship is shown in Figure 2.

Details of similar cases are not provided due to limited space in this paper.Interested readers are recommended to consult the example Lineage Modification file(File S2)for plausible lineage modifications.

On CVTree‘‘outliers”as compared with 16S rRNA taxonomy

Every classification depends on characters and criteria used. There is no a priori reason that 16S rRNA analysis and whole-genome approaches should yield identical results.The fact that they agree with each other in an overwhelming majority of cases confirms the objectivity of the present 16S rRNA-based taxonomy.However,minor discrepancies cannot be ignored,and these differences should be recorded and further studied.In addition to the aforementioned cases such as the phylum Spirochaetes,the class Deltaproteobacteria,the orders Myxococcales and Bdellovibrionales,and the‘‘genus”Coprothermobacter,we also include a few more as follows. (1)A new lineage from the genus Dehalococcoides to the class Dehalococcoidia was proposed recently within the phylum Chloroflexi[61].However,the 9 genomes from this taxon in CVTree,though forming a stable cluster,are not part of Chloroflexi and probably comprise a separate phylum.(2) Magnetococcus marinus was recently proposed to be a singlespecies lineage from genus to order at the base of the class Alphaproteobacteria[62].Although this species surely belongs to the phylum Proteobacteri a,it was separated from the main body of Alphaproteobacteria by a group of insect symbionts with highly-degenerated genomes.Whether this phenomenon is an artifact caused by the influence of their very small genomes[9,16]requires further analysis.(3)Hippea maritime definitely escapes from Deltaproteobacteria to the neighboring phylum Aquificae.

Figure 3 is a tree based on all3220 genomes,with the region of interest expanded to the rank of phylum and the rest collapsed as much as possible.Most points discussed above can be observed in this figure.Note that summing up the number of genomes shown explicitly in this figure yields 3219,because one genome used as out-group was hidden.

Infrasubspecific interrelationship within species

1.Serotypes of Streptococcus pyogenes.Figure 4 was isolated from a 3220-genome CVTree.The serotype ofthe 20 strains was placed in parentheses at the end of each entry,e.g.,M3 or M59.The branching order follows the serotype.

Figure 2 Candidate taxa at the phylum level near TenericutesThe CVTree3 server collapses candidate taxa at the phylum levelnear Tenericutes to a single note〈P〉Tenericutes{99+5}.The{n+m} notation indicates that there are n genomes with complete lineage information and m genomes with incomplete or missing lineage information.{n+m}is indicated as{n}when m=0,while when n=0,{n+m}is indicated as{0+m}.〈P〉,〈C〉,〈G〉,〈S〉,and〈T〉stand for phylum,class,genus,species,and strain,respectively.‘‘Unclassified”indicates missing classifier.Lineage information containing one or more‘‘Unclassified”is considered incomplete.

Figure 3 A 3220-genome CVTree collapsed to highlight the position of Acetothermia,Dehalococcoidia,Hippea,and Coprothermobacter The{n+m}notation indicates that there are n genomes with complete lineage information and m genomes with incomplete or missing lineage information.〈D〉and〈K〉,〈P〉,〈C〉,〈O〉,〈F〉,〈G〉,〈S〉,and〈T〉stand for domain,kingdom,phylum,class,order,family,genus, species,and strain,respectively.‘‘Unclassified”indicates missing classifier.Lineage information containing one or more‘‘Unclassified”is considered incomplete.The fraction 2577/2733 means 2577 from a totalof 2733 genomes.

Figure 4 Serotypes of the Streptococcus pyogenes strainsOnly strain tags are shown.Serotypes are given at the end of each entry.Whether the unknown{?}is M53 may be tested when serotype of the strain becomes available.〈T〉stands for strain.

2.Population genetics of bacteria is an important but much less-studied subject.For example,the clonal structure of naturally occurring Escherichia coli communities persists despite frequent recombination events[63,64]. Experimental methods are available for determining the phylogroups of E.coli,as the groups are associated with pathogenic or commensal behavior.Figure 5 shows a branch composed of 67 E.coli strains.The major branches in this figure correspond to the well-known phylogroups.

Figure 5 Phylogroups of 67 Escherichia coli strainsThe phylogroups A,B1,B2,D,and E are shown on the common branches.〈T〉stands for strain.

3.Biogeographic distribution of plants and animals lays the foundation for Darwin’s theory of evolution.However, the biogeography of bacteria has not been thoroughly examined.We refer to a recent paper[22]describing various strains of Sulfolobus islandicus collected from different parts of Euro-Asian and North-American continents as geovars.This work used CVTree and electronic DNA–DNA hybridization.The subtrees of Helicobacter pylori or Chlamydia trachomatis strains may be correlated with human migration patterns(figures not shown,but can be obtained from the Working Project of CVTree3).

4.As a potential application of CVTree’s high resolution power,electronic screening may be used to examine bacterial metabolic products.Screening bacterial mutant strains for pharmaceutical purposes is a costly and timeconsuming process.However,when a sufficiently large amount of experimental data has been accumulated,it is easy to map new mutant genomes into a phylogenetic tree based on previous screening knowledge labeled on the branches.Thus,only promising strains are selected for further examination in laboratory screening.

Discussion

Biology commences with taxonomy.However,the field of taxonomy as a subfield of biology,and particularly microbial taxonomy,is on the decline.Although the number of living microbial cells is estimated to be of the order of 1030[1] and the number of species surely exceeds 106[65],the number of described species is only slightly above 1.1×104[10].The pace of describing prokaryotic species will likely not catch up with the speed of discovery of new microorganisms.So called‘‘minimal standards”,practiced by some editorial offices of microbiological journals[66],further hinder the valid publication of bacterial names.In the 20-th century,the International Code of Nomenclature of Bacteria[57]has played a unifying role in the field of microbial taxonomy.However, as pointed out by Barny Whitman,the supervisor of the Bergey’s Manual,‘‘many biologists will no longer validate the names of newly described prokaryotes and the literature will once again be full of names with uncertain meaning”[66].Fortunately,development of genome sequencing technology could help provide a solution to such situation.‘‘With the availability of inexpensive DNA sequencing,prokaryotic species could be routinely described based upon their genome sequences”[66].Both phylogeny and taxonomy can become by-products of genomic analysis.Reliable and easily-usable tools such as CVTree3 will play a crucial role in future development.

The whole-genome approach will not replace other methods.In contrast,we advocate for the viewpoint of polyphasic phylogeny and taxonomy.With the cost of bacterial genome sequencing dropping below that of an average phenotyping experiment,the results of phenotyping tests have become even more valuable.A tripartite comparison between whole-genome based CVTree,16S rRNA sequences based All-Species Living Tree[67],and the Bergey’s Manual complemented by current taxonomic literature as reflected in LPSN[51]is becoming a feasible task.

Authors’contributions

BH initiated the CVTree approach.GZ designed and implemented the new parallelized web server.BH and GZ together tested the system and developed the biological applications of CVTree3 described in the paper.BH wrote the manuscript. Both authors read and approved the final manuscript.

Competing interests

The authors have declared that no competing interests exist.

Acknowledgments

This work was supported by the National Basic Research Program of the Ministry of Science and Technology of China (973 Project;Grant No.2013CB834100)and the State Key Laboratory of Applied Surface Physics as well as the Department of Physics,Fudan University,China.Dr.Wanzhen Zeng from Otawa University Hospital helped with collecting Streptococcus serotypes;Drs.Zhao Xu from Thermo Fisher Scientific and Xiaoyang Zhi from Yunnan University took part in discussions related to the new release of CVTree3.Prof.Jingchu Luo from Peking University made essentialsuggestions to improve the web server.The authors thank them all.

Supplementary material

Supplementary material associated with this article can be found,in the online version,at http://dx.doi.org/10.1016/j. gpb.2015.08.004.

References

[1]Whitman WB,Coleman DC,Wiebe WJ.Prokaryotes:the unseen majority.Proc Natl Acad Sci U S A 1998;95:6578–83.

[2]Fox GE,Pechman KR,Woese CR.Comparative cataloging of 16S ribosomal ribonucleic acid:molecular approach to procaryotic systematics.Int J Syst Bacteriol 1977;27:44–57.

[3]Bergey’s Manual Trust.Bergey’s manual of systematic bacteriology,2nd ed.,vol 1–5.New York:Springer-Verlag;2001–2012.

[4]Woese CR,Stackebrandt E,Macke TJ,Fox GE.A phylogenetic definition of the major eubacterial taxa.Syst Appl Microbiol 1985;6:143–51.

[5]Wayne LG,Brenner DJ,Colwell RR,Grimont PAD,Kandler O, Krichevsky MI,et al.Report of the ad hoc committee on reconciliation of approaches to bacterial systematics.Int J Syst Bacteriol 1987;37:463–4.

[6]Coneye T,Gevers D,Van der Peer Y,Vandamme P,Swings J. Towards a prokaryotic genomic taxonomy.FEMS Microbiol Rev 2005;29:147–67.

[7]Konstandinidis KT,Tiedje JM.Towards a genome-based taxonomy for prokaryotes.J Bacteriol 2005;187:6258–64.

[8]Klenk HP,Goker M.En route to a genome-based classification of Archaea and Bacteria.Syst Appl Microbiol 2010;33:175–82.

[9]Jun SR,Sims GE,Wu GA,Kim SH.Whole-proteome phylogeny of prokaryotes by feature frequency profiles:an alignment-free method with optimalfeature resolution.Proc NatlAcad Sci U S A 2010;107:133–8.

[10]Chun J,Rainey FA.Integrating genomics into the taxonomy and systematics ofthe Bacteria and Archaea.Int J Syst Evol Microbiol 2014;64:316–24.

[11]Wu D,Hugenholtz P,Mavromtis K,Pukall R,Dalin E,Ivanova NN,et al.A phylogeny-driven encyclopedia of Bacteria and Archaea.Nature 2009;462:1056–60.

[12]Kyrpides NC,Hugenholtz P,Eisen JA,Woyke T,Go¨ker M, Parker CT,et al.Genomic encyclopedia of Bacteria and Archaea: sequencing a myriad of type strains.PLoS Biol2014;12:e1001920.

[13]Qi J,Wang B,Hao BL.Whole genome prokaryote phylogeny without sequence alignment:a K-string composition vector approach.J Mol Evol 2004;58:1–11.

[14]Hao BL,Qi J.Prokaryote phylogeny withoutsequence alignment: from avoidance signature to composition distance.J Bioinform Comput Biol 2004;2:1–19.

[15]Gao L,Qi J,Sun JD,Hao BL.Prokaryote phylogeny meets taxonomy:an exhaustive comparison of composition vector trees with systematic bacteriology.SciChina C Life Sci2007;50:587–99.

[16]Li Q,Xu Z,Hao BL.Composition vector approach to whole genome-based prokaryote phylogeny:success and foundations.J Biotechnol 2010;149:115–9.

[17]Hao BL.Whole-genome based prokaryotic branches of the tree of life.In:Long M,Hongya G,Zhou Z,editors.Darwin’s heritage today:Proceedings of the Darwin 200 Beijing International Conference.Beijing:High Education Press;2010.p.101–13.

[18]Sun JD,Xu Z,Hao BL.Whole-genome based Archaea phylogeny and taxonomy–a composition vector approach.Chin Sci Bull 2010;55:2323–8.

[19]Zuo GH,Xu Z,Yu HJ,Hao BL.Jackknife and bootstrap tests of the composition vector trees.Genomics Proteomics Bioinformatics 2010;8:262–7.

[20]Hao BL.CVTrees support the Bergey’s systematics and provide high resolution at species level and below.Bull BISMiS 2011;2 (Part 2):189–96.

[21]Zuo GH,Xu Z,Hao BL.Shigella species are not strains of Escherichia coli but sister members in the genus Escherichia. Genomics Proteomics Bioinformatics 2013;11:61–5.

[22]Zuo GH,Hao BL,Staley JT.Geographic divergence of‘Sulfolobus islandicus’strains assessed by genomic analyses including electronic DNA hybridization confirms they are geovars.Antonie van Leeuwenhoek 2014;105:431–5.

[23]Zuo GH,Li Q,Hao BL.On K-peptide length in composition vector phylogeny of prokaryotes.Comput Biol Chem 2014;53:166–73.

[24]Zuo GH,Xu Z,Hao BL.Phylogeny and taxonomy of Archaea:a comparison of the whole-genome-based CVTree approach with 16S rRNA sequence analysis.Life 2015;5:949–68.

[25]Chan RH,Chan TH,Yeung HM,Wang RW.Composition vector method based on maximal entropy principle for sequence comparison.IEEE/ACM Trans Comput Biol Bioinform 2012;9: 79–87.

[26]Song K,Ren J,Reinert G,Deng MH,Waterman MS,Sun FZ. New developments of alignment-free sequence comparison:measures,statistics and next-generation sequencing.Brief Bioinform 2013;15:343–53.

[27]Qi J,Luo H,Hao BL.CVTree:a phylogenetic tree reconstruction toolbased on whole genomes.Nucleic Acids Res 2004;32:W45–7.

[28]Xu Z,Hao BL.CVTree update:a newly designed phylogenetic study platform using composition vectors and whole genomes. Nucleic Acids Res 2009;37:W174–8.

[29]Farris JS.Formal definitions of paraphyly and monophyly.Syst Zool 1974;23:548–54.

[30]Farris JS.Haeckel,history,and hull.Syst Zool 1990;39:81–8.

[31]Rinke C,Schwientek P,Sczybra A,Ivanova NN,Anderson IJ, Cheng J-F,et al.Insights into the phylogeny and coding potential of microbial dark matter.Nature 2013;499:431–7.

[32]Giongo A,Tyler HL,Zipperer UN,Triplett EW.Two genome sequences of the same bacterial strain,Gluconacetobacter diazotrophicus PIA 5,suggest a new standard in genome sequence submission.Stand Genomic Sci 2010;2:309–17.

[33]Gao L,Qi J,Wei HB,Sun YG,Hao BL.Molecular phylogeny of coronaviruses including human SARS-CoV.Chin Sci Bull 2003; 48:1170–4.

[34]Gao L,Qi J.Whole genome molecular phylogeny oflarge dsDNA viruses using composition vector method.BMC Evol Biol2007;7: 41.

[35]Chu KS,Qi J,Yu ZG,Ahn V.Origin and phylogeny of chloroplasts revealed by a simple correlation analysis of complete genomes.Mol Biol Evol 2004;28:70–6.

[36]Wang H,Xu Z,Gao L,Hao BL.A fungalphylogeny based on 82 complete genomes using the composition vector method.BMC Evol Biol 2009;9:195.

[37]Woese CR,Fox GE.Phylogenetic structure of the prokaryotic domain:the primary kingdoms.Proc Natl Acad Sci U S A 1977;74:5088–90.

[38]Garrity GM,Johnson KL,Lilburn TG.Taxonomic outline of the procaryotes.Bergey’s manualof systematic bacteriology,Rel.3.0. 2nd ed.New York:Springer;2002.http://dx.doi.org/10.1007/ bergeysoutline200210.

[39]Garrity GM,Bell JA,Lilburn TG.Taxonomic outline of the procaryotes.Bergey’s manualof systematic bacteriology,Rel.4.0. 2nd ed.New York:Springer;2003.http://dx.doi.org/10.1007/ bergeysoutline200310.

[40]Garrity GM,Bell JA,Lilburn TG.Taxonomic outline of the procaryotes.Bergey’s manualof systematic bacteriology,Rel.5.0. 2nd ed.New York:Springer;2004.http://dx.doi.org/10.1007/ bergeysoutline200405.

[41]TakaiK,Suzuki M,Nakagawa S,Miyazaki M,SuzukiY,Inagaki F,et al.Sulfurimonas paralvinellae sp.nov.,a novel mesophilic, hydrogen-and sulfur-oxidizing chemolithoautotroph within the Epsilonproteobacteria isolated from a deep-sea hydrothermalvent polychaete nest,reclassification of Thiomicrospira denitrificans as Sulfurimonas denitrificans comb.nov.and emended description of the genus Sulfurimonas.Int J Syst Evol Microbiol 2006;56: 1725–38.

[42]Fardeau M-L,Bonilla-Salinas M,L’Haridon S,Jeanthon C, Verhe F,Cayol J-L,et al.Isolation from oil reservoirs of novel thermophilic anaerobes phylogenetically related to Thermoanaerobacter subterraneus:reassignment of T.subterraneus,Thermoanaerobacter yonseiensis,Thermoanaerobacter tengcongensis and Carboxydibrachium pacificum to Caldanaerobacter subterraneus gen.nov.,sp.nov.,comb.nov.as four novelsubspecies.Int J Syst Evol Microbiol 2004;54:467–74.

[43]Hugenholtz P,Stackebrandt E.Reclassification of Sphaerobacter thermophiles from the subclass Sphaerobacteridae in the phylum Actinobacteria to the class Thermomicrobia in the phylum Chloroflexi.Int J Syst Evol Microbiol 2004;54:2045–51.

[44]Iverson C,Mullane N,McCardell B,Tall BD,Lehner A,Fanning S,et al.Cronobacter gen.nov.,a new genus to accommodate the biogroups of Enterobacter sakazakii,and proposalof Cronobacter sakazakii gen.nov.,comb.nov.,Cronobacter malonaticus sp.nov., Cronobacter turicensis sp.nov.,Cronobacter muytjensii sp.nov., Cronobacter genomospecies 1,and of three subspecies,Cronobacter dublinensis subsp.Dublinensis subsp.nov.,Cronobacter dublinensis subsp.lausannensis subsp.nov.and Cronobacter dublinensis subsp.lactaridi subsp.nov.Int J Syst Evol Microbiol 2008;56: 1442–7.

[45]Liu C,Finegold SM,Song Y,Lwson PA.Reclassification of Clostridium coccoides,Ruminococcus hansenii,Ruminococcus hydrogenotrophicus,Ruminococcus luti,Ruminococcus productus and Ruminococcus schinkii as Blautia coccoides gen.nov.,comb. nov.,Blautia hansenii comb.nov.,Blautia hydrogenotrophica comb.nov.,Blautia luti comb.nov.,Blautia productus comb.nov., Blautia schinkii comb.nov.,and description of Blautia wexlerae sp.nov.,isolated from human faeces.Int J Syst Evol Microbiol 2008;58:1896–902.

[46]Korczak BM,Kuhnert P.Phylogeny ofpasteurellaceae.In:Kuhnert P,Christensen H,editors.Pasteurellaceae biology,genomics and molecular aspects.Norwich,UK:Caister Academic;2008.p.27–52.

[47]Kuhnert P,Schalten E,Haefner S,Mayor D,Frey J.Basfia succiniciproducens gen.nov.,sp.nov.,a new member ofthe family Pasteurellaceae isolated from bovine rumen.Int J Syst Evol Microbiol 2010;60:44–50.

[48]Klenk H-P,Lapidus A,Chertkov O,Copeland A,Del Rio TG, Nolan M.Complete genome sequence of the thermophilic, hydrogen-oxidizing Bacillus tusciae type strain(T2T)and reclassification in the new genus,Kyrpidia gen.nov.as Kyrpidia tusciae comb.nov.and emendation of the family Alicyclobacillaceae da Costa and Rainey 2010.Stand Genomic Sci 2011;5:121–34.

[49]Kunisawa T.The phylogenetic placement ofhe non-phototrophic, Gram-positive thermophile‘Thermobaculum terrenum’and branching orders within the phylum‘Chloroflexi’inferred from gene order comparisons.Int J Syst Evol Microbiol 2011;61: 1944–53.

[50]Yutin N,Galperin MY.A genomic update on clostridial phylogeny:Gram-negative spora formers and other misplaced clostridia.Environ Microbiol 2013;15:2631–41.

[51]Parte AC.LPSN–List of prokaryotic names with standing in nomenclature.Nucleic Acids Res 2014;42:D613–6.

[52]Ramirez-Bahena M-H,Chahboune R,Peix A,Velazquer E. Reclassification of Agromonas oligotrophica into the genus Bradyrhizobium as Bradyrhizobium oligotrophicum comb.nov. Int J Syst Evol Microbiol 2013;63:1013–6.

[53]Chan PP,Cozen AE,Lowe TM.Reclassification of Thermoproteus neutrophilus Stetter and Zilling 1989 as Pyrobaculum neutrophilum comb.nov.based on phylogenetic analysis.Int J Syst Evol Microbiol 2013;63:751–64.

[54]Sen A,Daubin V,Abrouk D,Gifford I,Berry AM,Normand P. Phylogeny of the class Actinobacteria revisited in the light of complete genomes.The orders‘Frankiales’and Micrococcales’should be split into coherent entities:proposal of Frankiales ord. nov.,Geodermatophilales ord.nov.,Acidothermales ord.nov.and Nakamurellales ord nov.Int J Syst Evol Microbiol 2014;64: 3821–32.

[55]Gupta RS,Naushad S,Baker S.Phylogenetic analyses and molecular signatures for the class Halobacteria and its two major clades:a proposal for division of the class Halobaceria into an emended order Halobacteriales and two new orders,Haloferacales ord.nov.and Natrialbales ord.nov.Int J Syst Evol Microbiol 2015;65:1050–69.

[56]Cavalier-Smith T.The neomuran origin of archaebacteria,the negibacterial root of the universal tree and bacterial megaclassification.Int J Syst Evol Microbiol 2002;52:7–76.

[57]Lapage SP,Sneath PHA,Lessel EF,Skerman VBD,Seeliger HPR,Clark WA.International Code of Nomenclature of Bacteria.Bacterial Code 1990.Washington,DC:ASM Press; 1992.

[58]Bhatnagar S,Badar JH,Madupu R,Khouri HM,O’Oconnor Y, Robb FT,et al.Genome sequence of the sulfate-reducing thermophilic bacterium Thermodesulfovibrio yellowstonii strain DSM 11347T(phylum Nitrospirae).Genome Announc 2015;3: e01489–514.

[59]Collins MD,Lawson PA,Willems A,Cordoba JJ,Fernandez-Garayzabal J,Garcia P,et al.The phylogeny of the genus Clostridium:proposal of five new genera and eleven new species combinations.Int J Syst Bacteriol 1994;44:812–26.

[60]Schloss PD,Handelsman J.Status of the microbial census. Microbiol Mol Biol Rev 2004;68:686–91.

[61]Lo¨ffler FE,Yan J,Ritalahti KM,Adrian L,Edwards EA, Konstantinidis KT,Muller JA,et al.Dehalococcoides mccartyi gen.nov.,sp.nov.,obligately organohalide-respiring anaerobic bacteria relevantto halogen cycling and bioremediation,belong to a novel bacterial class,Dehalococcoidia classis nov.,order Dehalococcoidales ord.nov.and family Dehalococcoidaceae fam. nov.,within the phylum Chloroflexi.Int J Syst Evol Microbiol 2013;63:625–35.

[62]Bazylinski DA,Williams TJ,Lefevre CT,Berg RJ,Zhang CL, Bowser SS,et al.Magnetococcus marinus gen.nov.,sp.nov.,a marine,magnetotactic bacterium that represents a novel lineage (Magnetococcaceae fam.nov.,Magnetococcales ord.nov.)at the base of the Alphaproteobacteria.Int J Syst Evol Microbiol 2013;63:801–8.

[63]Selander RK,Caugant DA,Whitman TS.Genetic structure and variation in naturalpopulations of Escherichia coli.In:Niedhardt C,editor.Escherichia coli and Salmonella typhimurium.Cellular and molecular biology.Washington,DC:American Society for Microbiology;1987.p.1625–47.

[64]Tenaillon O,Skurnik D,Picard B,Denamur E.The population genetics of commensal Escherichia coli.Nature Rev Microbiol 2010;8:207–17.

[65]Curtis TP,Sloan WT,Scannell JW.Estimating prokaryotic diversity and its limits.Proc Natl Acad Sci U S A 2002;99: 10494–9.

[66]Whitman WB.Intent of the nomenclatural Code and recommendations about naming new species based on genomic sequences. Bulletin BISMiS 2011;2(Part 2):135–9.

[67]Yarza P,Richter M,Peplies J,Euzeby J,Amann R,Schleifer KH,et al.The All-Species Living Tree project:a 16S rRNA-based phylogenetic tree of all sequenced type strains.Syst Appl Microbiol 2008;31:241–50.

22 July 2015;accepted 10 August 2015

.

E-mail:hao@mail.itp.ac.cn(Hao B).aORCID:0000-0002-7822-5969.bORCID:0000-0003-3547-564X.

Peer review under responsibility of Beijing Institute of Genomics, Chinese Academy of Sciences and Genetics Society of China.