Reconstruction of an ancestral Yersinia pestis genome and comparison with an ancient sequence

13 Pages·2015·1.52 MB·English

Ducheminetal.BMCGenomics2015,16(Suppl10):S9
http://www.biomedcentral.com/1471-2164/16/S10/S9
RESEARCH Open Access
Yersinia pestis
Reconstruction of an ancestral
genome and comparison with an ancient sequence
Wandrille Duchemin1, Vincent Daubin1, Eric Tannier1,2*
From 13th Annual Research in Computational Molecular Biology (RECOMB) Satellite Workshop on
Comparative Genomics
Frankfurt, Germany. 4-7 October 2015
Abstract
Background: We propose the computational reconstruction of a whole bacterial ancestral genome at the
nucleotide scale, and its validation by a sequence of ancient DNA. This rare possibility is offered by an ancient
sequence of the late middle ages plague agent. It has been hypothesized to be ancestral to extant Yersinia pestis
strains based on the pattern of nucleotide substitutions. But the dynamics of indels, duplications, insertion
sequences and rearrangements has impacted all genomes much more than the substitution process, which makes
the ancestral reconstruction task challenging.
Results: We use a set of gene families from 13 Yersinia species, construct reconciled phylogenies for all of them,
and determine gene orders in ancestral species. Gene trees integrate information from the sequence, the species
tree and gene order. We reconstruct ancestral sequences for ancestral genic and intergenic regions, providing
nearly a complete genome sequence for the ancestor, containing a chromosome and three plasmids.
Conclusion: The comparison of the ancestral and ancient sequences provides a unique opportunity to assess the
quality of ancestral genome reconstruction methods. But the quality of the sequencing and assembly of the
ancient sequence can also be questioned by this comparison.
Background isolatedgenes,ancestralgenomereconstructionsarenow
Extantspeciesarederivedfromaprocessofevolutionand robustatascalelargerthanthegene,forfragmentswhere
diversificationfromspeciesnowdisappeared.Thesespe- no rearrangement have occurred [4]. Methods for infer-
ciesarecalledancientingeneralandancestraliftheylefta ring ancestral gene ordershave alsobeen explored [5-8].
descendant. Ancestral genomic sequences can be esti- Together,thesemethodsopenthewaytothereconstruc-
matedthroughcomputationfromasetofextantsequences tion of complete ancestral genomes, including their
relatedbyaphylogenyandamodelofevolution[1],while sequences.
ancientgenomicsequencesingeneralcanbesequenced Obtaining ancestral sequences can allow, through the
fromtheremainsofdeadorganisms[2]. study of physical properties of the reconstructed mole-
cules,theinferenceofthepaleoenvironnementsinwhich
Ancestral genome reconstruction these molecules evolved [9]. These methods also allow
Ancestralgenomereconstructioncanconsistinpredicting access to an oriented and ordered view of molecular
agenecontentinancestralspecies[3],and foreachgene eventsalongthehistoryoflife.Moreover,theyofferabet-
itssequence[1].Whileoriginallyusedtostudyproteinsor ter understanding of this history and can further our
knowledge of the mechanismslinking organic sequences
totheirfunctions[10].
*Correspondence:[email protected] Despite this, ancestral sequence reconstruction suffers
1LaboratoiredeBiométrieetBiologieÉvolutive,LBBE,UMRCNRS5558,
from several limits. Along with the study of molecular
UniversityofLyon1,43boulevarddu11novembre1918,69622
Villeurbanne,France evolution, it relies on the validity of models and their
Fulllistofauthorinformationisavailableattheendofthearticle
©2015Ducheminetal.;ThisisanOpenAccessarticledistributedunderthetermsoftheCreativeCommonsAttributionLicense
(http://creativecommons.org/licenses/by/4.0),whichpermitsunrestricteduse,distribution,andreproductioninanymedium,provided
theoriginalworkisproperlycited.TheCreativeCommonsPublicDomainDedicationwaiver(http://creativecommons.org/
publicdomain/zero/1.0/)appliestothedatamadeavailableinthisarticle,unlessotherwisestated.
Ducheminetal.BMCGenomics2015,16(Suppl10):S9 Page2of13
http://www.biomedcentral.com/1471-2164/16/S10/S9
fundamental hypothesis. Furthermore, given that we are separatingthemmaketheirancestoragoodcandidatefor
interested in a phenomenon often distant in time, it is an ancestral reconstructionincludingboth sequence and
at best difficult to obtain proofs validating proposed pre- gene organization along the chromosome and the plas-
dictions. Thus, the validation of ancestral reconstruction mids.Howeverdespitetheshortevolutionarytime,while
methods is often limited to robustness tests, or simula- substitutions are quite rare [2], there is a very active
tions that themselves rely on the validity of the models dynamicsofrearrangements,insertionsequencespropaga-
of evolution [1]. tion, duplications, copy number variation (see Figure 2),
whichmakestheproblemchallenging.
Ancient genome sequencing The late-medieval ancient genome, likely close to that
AncientDNAsequencesisanotherwaytohaveanaccess ancestor, offers a validation opportunity for the ancestral
tothepasthistoryoflivingorganisms.Undercertaincon- reconstruction method. We achieve here this recon-
ditionsitispossibletoobtaingeneticmaterialthroughthe struction and perform the comparison.
sequencingoftheremainsofan organism.AncientDNA Notethatasequenceofthesamegenomewasproposed
sequencingbeganinthemiddleofthe80swiththeclon- recentlybyRajaramanetal.[30],butwasnotissuedfrom
ingandsequencingoffragmentsofmitochondrialDNAin ancestralreconstruction.The contigs of theancientgen-
a museum specimen of Equus quagga, an extinct equine ome werescaffolded witha methodincluding the phylo-
species that disappeared in the XIXth century [11]. The genyofrelatives,andsomepartsoftheassemblycouldbe
adventofPCRmethods[12]andhigh-throughputsequen- corrected,butwhatwepresenthereisnotusingatallthe
cing [13] followed by what is called third generation ancient sequence in the reconstruction phase, it is done
sequencing[14]allowedthesequencingofseveralextinct onlyfromindependentextantdata.
animals [15-17], ancient unicellular eukaryotes [18,19],
bacteria[2,20,21],metagenome[22],orvirome[23]. Methods
The ancient sequences disclose a new source of infor- An overview of the method, including species tree con-
mation concerning the evolution of lineages of interest. struction, gene tree construction and reconciliation,
They have already been used, among other things, to gene order inference and gene tree corrections accord-
understand the dynamic of extant populations of the ing to this gene order, and eventually genic and inter-
genus Homo [24-26], or other animals [27], to correct genic sequence prediction, is illustrated on Figure 3.
and recalibrate phylogenies [17], or to better understand
past pandemics [18,19,2,20,21]. Data set
However, along with the problems specific to sequen- The data consists in 13 Yersinia annotated genomes
cing technologies, ancient DNA sequencing islimited by (Figure 1) from which we extract 3772 homologouspro-
thepost-mortemchemicaldegradationofDNAmolecules teingenefamiliescontainingatleasttwogenes,usingthe
throughouttime.Thus,likefossils,ancientsequencesare HOGENOM database [31]. Of these, 1971 have exactly
scarcewhile,unlikethem,limitedtorecenttimes. onecopyperextantstrain.Thisstepcorrespondstopart
AinFigure3.
Yersinia pestis
ClassifiedamongEnterobacteriaceae,Yersiniapestisisthe Species tree
bacteriumthoughttoberesponsibleforthebubonicpla- Using Muscle [32] (default parameters), we aligned the
gue and the pneumonic plague. It diverged from the 1971families,concatenatedthevariablesitesofallalign-
Yersinia pseudotuberculosis lineage, in part through the mentsandobtainedaphylogenetictreeusingPhyML[33]
acquisitionoftwoplasmids[28].Ithasbeendemonstrated (100 bootstraps, otherwise default parameters) that we
that strains of Yersinia pestis caused the black death of rootedbyseparatingthepestisfromthepseudotuberculosis
1347-1353 AD that is thought to have killed between a clades, according to a consensus in the literature. In our
thirdandhalfoftheEuropeanpopulationatthattimeand tree the branch separating the two clades is well sup-
persistedinEuropeuntilthemiddleoftheXVIIIthcentury ported,aswell as the branchessurroundingthe ancestor
[29]. An ancient genome has been extracted and thatwewishtoreconstruct(seeFigure1).Thisstepcorre-
sequenced[2].Itwasthefirstwholeancientbacterialgen- spondstopartBinFigure3.
ome.Basedonasubstitutionpatterncomparedtoextant
Yersiniaspecies,ithasbeenhypothesizedtotakeplaceon Gene trees
the extant species phylogeny in the vicinity of a known AllgenefamiliessequenceswerethenalignedusingPrank
speciation node leading to two set of extant, sequenced [34] and one gene tree per family was computed using
andannotatedstrainsofthebacterium(seeFigure1). PhyML (100 bootstraps, otherwise default parameters).
Theexistenceofseveralsequencedandannotatedextant Because we are aligning recently diverged strains of the
genomes as well as the relatively short evolutionary time same organisms [35], the sequences often have not
Ducheminetal.BMCGenomics2015,16(Suppl10):S9 Page3of13
http://www.biomedcentral.com/1471-2164/16/S10/S9
Figure1Yersiniapestisandpseudotuberculosisphylogeny.Treeobtainedusinga1971universalgenefamiliesconcatenate.Bootstrap
valuesarefiguredonthebranches.Forreadability,thefiguredbranchlengthistheinverseoftheten-logarithmoftherealbranch-length.The
ancestralspeciesofinteresttousisfiguredasareddiamond.Thelatemedievalancientgenomehypotheticalpositionisfiguredingrayand
dashed.
Figure2DotplotbetweenthesequenceoftwoextantstrainsofYersiniapestis:CO92andKM10.Bothstrainsaredescendantsofthe
ancestorwefocuson.DatawasobtainedbyaligningthesequenceofstrainKIM10onthesequenceofstrainCO92usingmegablast(default
parameters,onlyhitswithalength>102.5werekept).
Ducheminetal.BMCGenomics2015,16(Suppl10):S9 Page4of13
http://www.biomedcentral.com/1471-2164/16/S10/S9
Figure3ProtocolusedtoobtaintheancestralgeneorderandsequenceofaYersiniapestisancestor.A)Extractionandfilteringofgene
familiesfromextantgenomesandalignment.B)Reconstructionofthespeciestreeusingaconcatenateofthevariantpositionsof1971universal
genefamilies.C)MLreconstructionofgenetreesfollowedbythecollapseofanynon-supportedbranch(bootstrap<99)andtheresolutionof
thecreatedpolytomiesusingthespeciestreeasaguide.D)InferenceofancestralgeneadjacenciesusingDeCo.E)Detectionandcorrectionof
wronglyinferredgenetreesbasedontheancestraladjacencygraphlinearity.F)Reconstructionoftheancestralsequenceofeachgene
adjacencyfromtheirextantdescendants.G)Alignmentoftheconsecutiveancestraladjacencysequencestoassembletheancestralgenome.
Similarcolorsindicateshomology.Dotsrepresentageneasanodeinanadjacencygraphwhileorientedsegmentsrepresentageneasa
sequence.
Ducheminetal.BMCGenomics2015,16(Suppl10):S9 Page5of13
http://www.biomedcentral.com/1471-2164/16/S10/S9
diverged enough to allow an unambiguous tree recon- betweenA′andB.Soextantgenomesaresetsofdisjoints
struction. So we collapsed all branches with a support cyclesinagraph,modelingchromosomesandplasmids.
lower than 99 and then used ProfileNJ [36] to solve the Gene extremities can be clustered into families, inher-
created polytomies. ProfileNJ reconstructs species tree ited from gene families, and also inherit the reconciled
branches instead of collapsed branches and chooses genetree.
amongseveralsolutionswithaNeighbor-Joiningformula.
Distances for the Neighbor-Joining part were computed Ancestral gene order
with bppdist, a Bio++ suite software [37] (GTR + Γ(4) Ancestral adjacencies between gene extremities were
model). inferred using DeCo [7]. It models the evolution of an
ProfileNJ also roots the gene trees according to “Last adjacencybetweentwogeneextremitiesfollowingaparsi-
Common Ancestor” reconciliation method, annotating mony principle, i.e. minimizingthe numberofgainsand
internal nodes with duplications or speciations, and breakages ofadjacencies,duetorearrangements. It takes
choosing a root minimizing the number of duplications. asinputthespeciestree,allgenetrees,andextantadjacen-
Reconciled gene trees depict the history of the gene cies,and proposesa set ofancestral adjacencies between
family, including all ancestral genes, uniquely defined by ancestralgeneextremitiesdefinedbythereconciledgene
the reconciliation. trees.ThisstepcorrespondstopartDinFigure3.
This step corresponds to part C in Figure 3. DeCo assumes that adjacencies evolve independently.
This means in particular that ancestral gene extremities
Gene families filtering can be involved in an arbitrary number of adjacencies.
From the 3772 gene families, some were discarded Ancestralgeneextremitiesandadjacenciesarenotneces-
because they showed signal of a process that we do not sarily made of cycles as extant genomes, so we call this
handle well in our pipeline, gene transfer. Transfer was object an adjacency graph. Figure 4 shows the obtained
suspected when a branch in the reconciled gene tree adjacencygraphatthisstep.Whilemostofitshowsalin-
would correspond to at least 4 independent losses in the ear orcircularstructure,there aresomegeneextremities
species tree. We also removed the families with more withtoomanyadjacencies,otherswithnotenough.
than 5 genes in the black death ancestor, suspecting Therecanbeseveralreasonsfortheadjacencygraphnot
insertion sequences, which are poorly handled by the tobeacollectionofpathsandcycles,aswewouldexpect
method. We also removed families containing genes ifthedataandmethodswereperfect.Incorrectgenetrees
fully included in other genes: as we model the evolution
of gene orders, these would be difficult to handle. We
eventually removed families when the reconciled gene
tree did not contain a gene in the ancestor we want to
reconstruct.
The final data set contained 3656 families. Note that
when removing gene families from the study, we do not
necessarily give up the reconstruction of parts of the
ancestral sequence. We just define the removed parts as
intergenic. As we also reconstruct intergenic sequences,
this simply modifies the resolution at which we are able
to detect rearrangements.
Extant gene order and adjacencies
Eachgeneisasegmentofachromosomeoraplasmidand
has a start and an end position on it. We identify these
positions as the extremities of the gene. A start position
may be greater than an end position: the order of the
extremitiesdefinestheorientationofthegene.Wemodel
each genome by a graph, whose nodesare gene extremi-
ties of genes in that genome. We put an edge, called an
Figure4 Ancestral adjacencygraph obtainedusing DeCo on
adjacency between pairs of extremities of a same gene.
thesetof3656genefamilies.Eachnodeiscoloredaccordingto
AdditionallyifgenesAA′andBB′areconsecutive(Aand itsnumberofneighbors:greenfortwo(ideal,linearcase),turquoise
A′ are the extremities of the first gene, appearing in that forone(whereoneadjacencyhasbeenlost),orangeforthreeand
order on the chromosome or plasmid, and B, B′ are the grayforfour(whenanerrorinthenumberofancestralcopies
createsconflictintheancestralgeneorder).
extremities of the second gene), we put an adjacency
Ducheminetal.BMCGenomics2015,16(Suppl10):S9 Page6of13
http://www.biomedcentral.com/1471-2164/16/S10/S9
areprobablythemajorsourceofsuchdiscrepancies,while turnitintoaspeciationnodebygivingittwodescendant
othersmaycomefrom uncertaintiesinadjacency history nodes h and h , and assigning its descendant leaves to
1 2
inference. eitheroneofthem,dependingonwhethertheyaregenes
Wetransformtheadjacencygraphintoagenome(i.e.an from descendants of s1or s . Then subtrees rootedath
2 1
adjacency graph that is a collection of paths and cycles), andh arereconstructedusingProfileNJ.
2
firstbycorrectinggenetrees,byoperationswecallzipping Figure 5B gives an example of a zipping operation on
andunzipping,then byremovingaminimumnumber of the ancestral adjacency graph and on the gene tree.
adjacenciessothattheremaininggraphisagenome. Zippingproducesanewancestralgeneh insteadoftwo
d
paralogueshandh′.Wepropagatethesameoperationto
Correcting gene trees the neighbors of the ancestral gene h in the adjacency
d
This step corresponds to part E in Figure 3 and a more graphiftheyarethemselvessupernumeraryparalogues.
detailed picture is on Figure 5. Note that for zipping and unzipping, the propagation
Unzipping mechanism allows the treatment of several consecutive
Each ancestral gene extremity of a gene g should have at nodes,suchthatalargesegmentalduplicationcontaining
most two adjacencies. If one has more than two, a first multiple genes can bedealtwith as long asthere existsa
hypothesis can be that in the real ancestral genome, the nodetostarttheunzippingmove(e.g.atoneextremityof
gene g was duplicated in two copies, and each copy thesegmentalduplication).
would carry some of the adjacencies of g. Cutting
If in one extant species, there are two homologous Zipping and unzipping are tested independently for each
copies of the gene g, and their extremities share the ancestral node with more than two neighbors. Each of
homologs of the adjacencies attributed to an extremity them should decrease the number of gene extremities
of g, then we perform the unzipping operation. with more than two adjacencies. The operation that
It consists in making two genes out of g by modifying decreases it the most is kept.
the gene tree T of the gene family containing g. Only If none of zipping and unzipping succeeds in removing
the subtree rooted at g is changed, into a subtree rooted all such supernumerary adjacencies (it is possible that
at a new duplication node with two descendants: g and none of the hypotheses applies), then we remove as few
a new gene g′. Then the two subtrees rooted at g and g′ adjacencies as possible so that only gene extremities with
are reconstructed, first by assigning all leaves to g or g′ at most two adjacencies remain. This is achieved using a
according to their neighborhood; Then by constructing maximum matching technique described in [39].
subtrees on these leaves using ProfileNJ. In the case
where some leaves can’t be assigned to either g or g′ Ancestral sequence reconstruction
using their neighborhood (i.e. their extant neighbors are Ancestral sequences have to be reconstructed by pieces,
not descendant of any of the ancestral neighbors), then because they need a multiple alignment free of rearran-
leaves are assigned to one of the two set of leaves gements. The pieces have to be glued together, and in
according to their mean phylogenetic distances with order to avoid between pieces border problems, pieces
them. Where there is a tie (for instance if all sequences have to overlap. This is why we reconstruct an ancestral
are identical, all distances are null), the leaf is randomly sequence for all pairs of genes which are connected by
assigned to one of the two leaf-set. an adjacency. Then pairs are aligned together on their
Figure 5A gives an example of an unzipping operation common gene, and merged.
on the ancestral adjacency graph and on the gene tree. Weorienteachadjacentgenepairwithafirstandasec-
If the unzipping procedure increases the number of ond gene, each gene should be once the first gene of a
adjacenciesincidenttoageneextremityofagenehinthe pair, and once the second in another pair. We use the
immediateneighborhoodofgintheadjacencygraph,then genetreeofthefirstgeneasaguide,toconstructamulti-
theunzippingprocedureisappliedtohaswell,andthen ple sequence alignment with the extant sequences that
toitsneighbors,untiltheregionislinearized. contain this adjacent pair (thus, the sequences contains
Zipping bothgenesandthesequencebetweenthemwhentheyare
Another possible reason for a gene g to be involved in neighborsin an extant species, andonly the first gene of
morethantwoadjacenciesisthattwooftheseadjacencies the adjacency when they aren’t), and the ancestral
ghandgh′concerntwoparalogshandh′whichinreality sequenceusingPrank[34].
should form only one gene. In that case we perform a Genesequencesattheendsofcontigsarereconstructed
zippingoperation,similartotheonedescribedin[38]. alone using their own tree. In consequence each inter-
Leth bethelastcommonancestor ofhandh′intheir gene sequence is reconstructed once and each gene
d
gene tree. Suppose itisassigned to speciess, whose des- sequenceisreconstructedtwiceandatleastoncewithits
cendants are s and s . It is a duplication node, and we own tree.We assembletheobtainedancestral sequences
1 2
Ducheminetal.BMCGenomics2015,16(Suppl10):S9 Page7of13
http://www.biomedcentral.com/1471-2164/16/S10/S9
Figure5Illustrationoftheunzippingandzippingongenetreesandadjacencygraphs.A)Priortolinearization(leftoftheblackarrow),the
genegexistsinonecopyintheancestor(verticalgraylineonthetree)andtwoindependentduplicationsoccursinitsdescendants(greenhollow
squares).Intheancestraladjacencygraphaboveeachofgextremitiesdisplaystwoneighbors.Unzipping(rightoftheblackarrow)modifiesthetree
sothattherearetwoancestralcopiesgandg′eachcorrespondingtoadifferentpathintheancestraladjacencygraph(lossesinthetreeare
displayedasredcrosses).B)Priortolinearization(leftoftheblackarrow),twoancestralcopiesofthesamegenehandh′existintheancestor(vertical
graylineonthetree;lossesinthetreearedisplayedasredcrosses).Intheancestraladjacencygraphabovetheextremitiesofhandh′eachsharea
neighbor,forminganon-linearpattern.Zipping(rightoftheblackarrow)modifiesthetreesothatthereonlyoneancestralcopyh followedby
d
independentduplicationsinitsdescendants(greenhollowedsquares),formingonelinearpathintheancestraladjacencygraph.
Ducheminetal.BMCGenomics2015,16(Suppl10):S9 Page8of13
http://www.biomedcentral.com/1471-2164/16/S10/S9
by aligning (using Smith & Waterman’s algorithm) the genome and two extant ones, while a comparison
ones sharing a gene and then making the consensus between the two extant ones is shown on Figure 6A. The
sequenceofthatalignment,favoringthesequencerecon- isolated dots on the dotplots of Figure 6B and C are
structedwiththetreeofthealignedgene. probably reconstruction errors. While they could be
For instance, consider the ancestral path ABC (where explained as small rearrangements, they probably are arti-
A,B and C are genes), we reconstruct the ancestral facts of the adjacency graph linearization method, like a
sequence of A using its own tree, AB using A’s tree, BC leaf falsely associated to a subtree in an unzipping event
using B’s tree and C using its own tree. Afterward the for instance.
ancestral sequence of A is aligned with the ancestral TheancestralsequencesoftheplasmidspCD,pMTand
sequence AB, favoring the sequence of A when comput- pPCP were entirely reconstructed, for a total of respec-
ing the consensus. Then the sequence AB is aligned tively100.1kb,67.7kband9.6kb.Concerningtheances-
with the sequence BC, favoring the sequence BC in the tralchromosome,atotalof4.7Mbofancestral sequence
consensus (as both sequences align on gene B and BC wasreconstructed,whichisclosetothesizeoftheextant
used B’s tree for the reconstruction). Finally, the chromosomesofYersiniapestisstrains(e.g.4.7Mbforthe
sequence ABC is aligned with the sequence C, favoring strainAntiqua).Alackofsignalinextantgenomesdueto
C in the consensus. convergentrearrangements,preventedthereconstruction
A graphical view of these steps are given in Figure 3, offourancestral adjacencies.Becauseofthese,theances-
parts F and G. tral chromosome sequence is actually composed of four
Note that, as stated before, the ancestral sequence disjoint fragments (their sizes are respectively 3.44 Mb,
reconstructionneedsamultiplealignmentfreeofrearran- 0.67Mb,0.40Mband0.19Mb).
gements. This means that the size of the recombination The reconstructed ancestral sequences are avalaible in
events that can be taken into account for ancestral Additional file 1.
sequences reconstruction depends on the density of the
markers(here,thegeneextremitiesof3656genefamilies) Comparison to the ancient genome
usedintheancestralorderreconstructionstep. UsingMegablast[41]wealignedthe2134ancientYersinia
pestiscontigsobtainedbyBosetal.[2](avalaibleathttp://
Results paleogenomics.irmacs.sfu.ca/FPSAC/,lastaccessed19june
The shape of the ancestral genome 2015) against the obtained ancestral genome, including
We perform the whole process of ancestral gene order chromosomeandplasmids.
reconstructionforthreedatasets:thewholesetoffiltered We examine 2179 hits of length >102.5bp from 2087
families,thesetofDfreefamilies,withoutduplicationand contigs(seeAdditionalfile2forthebimodaldistribution
theDLfreefamilies,withoutduplicationnorloss. of hit lengths which justifies this threshold). The others
Ancestral gene order is computed with the whole set, arefullofrepeatedelements,makingthecomparisondiffi-
butitgivesfragmentedpathsintheadjacencygraph.The cult.Asaconsequencetheexaminedhitsallmatchtothe
fragments are progressively assembled using the D free chromosomeandnonetotheplasmids.
andDLfreegeneorders. Gene order
The ancestral gene order was reconstructed for the These hits show a quasi-total congruence between the
chromosome (3342 genes) and the three plasmids (pCD: organization of the ancient and ancestral sequence.
74 genes, pMT: 87 genes, pPCP: 5 genes). The plasmids Figure 7representsthecorrespondence between thetwo
pCD and pPCP were obtained as circular elements in intheformofadotplot,wherecontigsoftheancientgen-
the adjacency graph, while the plasmid pMT was repre- omeareconcatenatedaccordingtotheancestralsequence.
sented by one linear fragment. The chromosome was Three isolated dotsdeviate fromthe central line. Two of
obtainedas three linearcomponents. Tojointhesecom- them concern large repeated regions, that is, the whole
ponents,weranDeCoontheirsixextremitiesusingagra- contigs match at several places. Only one seems to be a
dientofadjacencygain/losscostsratio(from1/10to10/1) realdiscordancebetweenthetwogenomes.Twocontigu-
and scored each potential adjacency by the number of ous regions of the contig hit on two different ancestral
timesitwasobserved.Wethenappliedaweightedmaxi- sequence fragments. This chimeric contig (number 8335
mummatchingtechnique[40]toextractthebestpossible in[2])hadalreadybeenobservedbyRajaramanetal.[30]
order between the fragments (only one optimal solution intheirscaffoldingoftheancient genome.Thisstretches
remained). the proximity and the differences between the two
The ancestral gene order is different from all extant approaches. Indeed, the latter, called FPSAC, takes as
genomes. For example it is an intermediary between the input the ancient contigs and the extant genomes, frag-
two extant strains CO92 and KIM10. Figure 6 B and C mentsthecontigsaccordingtotheiralignmentstoextant
show the gene order comparison between the ancestral genomes, and orders fragments. Here we don’t use at all
Ducheminetal.BMCGenomics2015,16(Suppl10):S9 Page9of13
http://www.biomedcentral.com/1471-2164/16/S10/S9
Figure 6 Dotplot between the ancestral genome and two extant strains of Yersinia pestis: CO92 and KIM10. Both strains are
descendantsoftheancestorwefocuson.DatawasobtainedusingtheextantadjacencygraphsofstrainsKIM10andCO92andconcernsgenes
order.Verticalandhorizontallinesseparatethedifferentmolecules(herethechromosomeandtheplasmids).A)dotplotbetweenthegene
ordersofthetwoextantstrainsKIM10andCO92.B)dotplotbetweenthegeneordersoftheancestralgenomeandtheextantstrainCO92.C)
dotplotbetweenthegeneordersoftheancestralgenomeandtheextantstrainKIM10.
theancientcontigsandstartfromextantgenes.Soweare case there is no comparison point to infer some
independentoftheextractionandassemblymethodology bases, and some are inferred differently than in the
fortheancientsequence,andwecancomparetoit.More- ancient sequence.
over,alloursequencesarecomputationallyreconstructed, (cid:129) Lack of a good model of evolution at an intermedi-
whichwasnotthecaseofthoseobtainedwithFPSAC. ary scale, like duplication of small elements. They
So at a large scale, there is only one difference which are here included in alignments and indel models,
can be an assembly error in the ancient sequence or a which do not account for repetitions.
derived mutation of the ancient bacteria, because the (cid:129) Assembly errors in the ancient sequence.
ancient configuration is not supported by extant
genomes. Consider for example the ancient contig number 497
Sequences where a mismatch occurs when aligned with the ances-
At a finer scale, differences are more numerous. tral sequence. The mismatch is situated in an intergenic
Approximately 81% of the 2084 contigs with a hit are region of the ancestral genome that is present in one
exact matches to the ancestral genome. We examined descendant of the reconstructed ancestor and two out-
some of the remaining and found that the differences group Yersinia pestis species. Consequently, the ances-
could be explained by three kinds of error sources in tral sequence was reconstructed using a tree where the
the ancestral or ancient sequences: node of interest was along a branch, missing a compari-
son point (i.e. another descendant) to choose between
(cid:129) Lack of sufficient data for ancestral reconstruction: its descendant allele and the outgroup allele.
it is the case if only one of the two children which Consider also the ancient contig number 8849 which
branches off the ancestor, in addition to an out- aligns with one mismatch to the reconstructed ancestor.
group, support the presence of a sequence. In that At the position of the mismatch, all extant (group and
Ducheminetal.BMCGenomics2015,16(Suppl10):S9 Page10of13
http://www.biomedcentral.com/1471-2164/16/S10/S9
chimeric contig but this discrepancy is independent).
Around position 1860, that ancient contig displays one
occurrence of a 20-mer. However, the reconstructed
ancestral sequence has two consecutive occurrences of
that 20-mer. This region is situated in an intergenic
region, so it has been reconstructed by an alignment of
an adjacency with its two flanking genes. The extant
species (descendant of the reconstructed ancestor or
not) which have this gene adjacency all display two
occurrences (in favor of the ancestral reconstruction) at
the exception of Yersinia pestis strain CO92, the Yersi-
nia pestis reference genome which was used to map the
ancient reads in [2]. While the fact that we did not use
the raw reads obtained in [2] prevents us to draw any
definitive conclusion, this appears to be an error in the
ancient sequence assembly, caused by a derived muta-
tion in the genome used as a reference.
Conversely, it happens that similar patterns are better
explained by errors in the reconstructed ancestral
sequence. Such a case occurs on the locus where the
ancient contig number 5613 maps. The situation is also
Figure 7 Dotplot between the late medieval Yersinia pestis
similartoFigure8A.Twocontiguousregionshitatadis-
genomesandthereconstructedancestralsequence.The
reconstructedchromosomewasalignedtothe2134ancient tanceof1315bponthereconstructedancestralsequence.
contigs,usingmegablast(defaultparameters,onlyhitswitha The sequence separating the two hits in the ancestor is
length>102.5werekept).Contigsareconcatenatedaccordingto only supported by one extant descendant (Nepal strain)
thereconstructedsequence,sotheagreementispartlyduetothe
andtheotherextantdescendantsmatchtheancientcontig
fragmentednatureoftheancientsequence.Thecontigswithhits
inonlyonelonghit.Thisseemstobeanerrorduetothe
departingfromthediagonalarecircledinred.
absenceofanevolutionarymodelallowingbiginsertions.
Prankmodelsindelsbut1315bpisnotreallyanindelbut
outgroup species) sequences bear the same allele and isratheraninsertionofwhatshouldperhapshavebeenan
thus the reconstructed ancestral sequence bears it too. evolutionary unit. It seems that the indel model prefers
However, the ancient contig bears another allele at that losingseveraltimessuchalongDNAsegmentratherthan
position. If we consider the ancient contig as correct, insertingitonceinaterminalbranchofthephylogeny.So
then this difference would be an original mutation on we can expect a small number of such false additions in
the ancient strain. Such an hypothesis could be checked theancestralsequence.
by mapping the ancient reads to their contigs in order
to assess the validity of that specific allele. However, we Discussion
note that the original study [2] that used read data to A complete reconstruction of an ancestral genome at
call SNPs did not detect any that were specific to the the nucleotide level requires to take into account evolu-
ancient strain. tionary events at several scales: nucleotide substitutions,
There are also differences that are more structural in indels, duplications, losses, recombinations, transfers,
kind. For example 43 contigs show some structural dif- transposable elements propagation, rearrangements.
ferences with the ancestral genome. On 39 of them, the Each level is handled by dedicated bioinformatics tools
ancient contig displays two contiguous or slightly over- which are rarely used together.
lapping hits that are more distant on the ancestral gen- We associated here gene content/sequence/order tools
ome (on 21 occasions, they are more than 300 bp apart inordertoattemptthereconstructionofawholeancestral
in the ancestral sequence), as in Figure 8A. On 4 ancient bacterialgenome,includingachromosomeandthreeplas-
contigs, contiguous regions are shown as overlapping in mids. We chose an organism from the Yersinia pestis
the ancestral genomes, as in Figure 8B. clade because of a recently published ancient sequence.
Such discrepancies can sometimes be explained by Despite being relatively recent at the evolutionary scale
errors in the ancient sequence, especially in regions (650years),theevolutionatalllevels,andinparticularin
where repetitions occur. For instance, the case illu- genome structure and organization, makes the problem
strated on Figure 8A, is seen on the contig number difficult. The difficulty can come from numerous events
8335 obtained by Bos et al.[2] (which is also the (rearrangements, insertion sequence dynamics), but also