Joint assembly and genetic mapping of the Atlantic horseshoe crab genome reveals ancient whole genome duplication

PAGE 1

RESEARCHOpenAccessJointassemblyandgeneticmappingofthe Atlantichorseshoecrabgenomerevealsancient wholegenomeduplicationCarlosWNossa1,4,PaulHavlak1,Jia-XingYue1,JieLv1,KimberlyYVincent1,HJaneBrockmann3andNicholasHPutnam1,2*AbstractBackground: Horseshoecrabsaremarinearthropodswithafossilrecordextendingbackapproximately450million years.Theyexhibitremarkablemorphologicalstabilityovertheirlongevolutionaryhistory,retaininganumberof ancestralarthropodtraits,andareoftencitedasexamplesof “ livingfossils. ” Asarthropods,theybelongtothe Ecdysozoa ,anancientsuper-phylumwhosesequencedgenomes(includinginsectsandnematodes)havethusfar shownmoredivergencefromtheancestralpatternofeumeta zoangenomeorganizationthancnidarians,deuterostomes andlophotrochozoans.However,muchofecdysozoandiversity remainsunrepresent edincomparativegenomicanalyses. Results: Hereweapplyanewstrategyofcombined denovo assemblyandgeneticmappingtoexaminethe chromosome-scalegenomeorganizationoftheAtlantichorseshoecrab, Limuluspolyphemus .Weconstructedagenetic linkagemapofthis2.7GbpgenomebysequencingthenuclearD NAof34wild-collected,ful l-siblingembryosandtheir parentsatameanredundancyof1.1xpersample.Themapincludes84,307sequencemarkersgroupedinto1,876distinct geneticintervalsand5,775candida teconservedproteincodinggenes. Conclusions: Comparisonwithothermetaz oangenomesshowsthatthe L.polyphemus genomepreservesancestral bilaterianlinkagegroups,andthatacommonancestoro fmodernhorseshoecrabsunderwentoneormoreancient wholegenomeduplications300millionyearsago,followed byextensivechromosomefusion.Theseresultsprovidea counter-exampletotheoftennotedcorrelationbetweenwh olegenomeduplicationandev olutionaryradiations.The new,low-costgeneticmappingmethodforobtainingachr omosome-scaleviewofnon-modelorganismgenomesthat wedemonstrateheredoesnotrequirelaboratoryculture,and ispotentiallyapplicabletoabroadrangeofotherspecies. Keywords: Genotyping-by-sequencing(GBS),Geneticlinka gemapping,Genomeevolution,LimuluspolyphemusBackgroundComparativeanalysisofgenomesequencesfromdiverse metazoanshasrevealedmuchabouttheirevolutionover hundredsofmillionsofyears.Thediscoveryofextensive genehomologyacrosslargeevolutionarydistanceshas allowedresearcherstotrack chromosomerearrangements andwholegenomeduplications.Theresultingvalueof wholechromosomesequencespresentsachallengefor existingwholegenomeshotgun (WGS)assemblystrategies. Wholegenomeduplicationeventswerelongsuspected [1],butonlytheavailabilityofgenomesequenceshas allowedconfirmationoftheminfungal,vertebrate,plant andciliatelineages[2-5].Incontrast,whenonlyafew chordate,insectandnematodegenomeswereavailable, conservationofgenelinkage(i.e.,synteny)andgene orderwereobservedonlybetweenclosely-relatedspecies, andconsequentlywerenotexpectedtobeconserved betweenphyla.Asmoremetazoangenomeshavebeensequenced,ithasbecomeclearthatlong-rangelinkagehas beenconservedoverlongtimescalesinmanylineages. Sequencingthegenomesofrepresentativesofchordate, mollusk,annelid,cnidarian,placozoanandspongeclades, hasidentified17or18ancestrallinkagegroups(ALGs) *Correspondence: nputnam@gmail.com1DepartmentofEcologyandEvolutionaryBiology,RiceUniversity,P.O.Box 1892,Houston,TX77251-1892,USA2DepartmentofBiochemistryandCellBiology,RiceUniversity,P.O.Box1892, Houston,TX77251-1892,USA Fulllistofauthorinformationisavailableattheendofthearticle 2014Nossaetal.;licenseeBioMedCentralLtd.ThisisanOpenAccessarticledistributedunderthetermsoftheCreative CommonsAttributionLicense(http://creativecommons.org/licenses/by/2.0),whichpermitsunrestricteduse,distribution,and reproductioninanymedium,providedtheoriginalworkisproperlycredited.TheCreativeCommonsPublicDomain Dedicationwaiver(http://creativecommons.org/publicdomain/zero/1.0/)appliestothedatamadeavailableinthisarticle, unlessotherwisestated.Nossa etal.GigaScience 2014, 3 :9 http://www.gigasciencejournal.com/content/3/1/9

PAGE 2

[6-10].EachoftheseALGsconsistsofasetofancestral geneswhosedescendantsshareconservedsyntenyin multiplesequencedgenomes.TheseALGshavebeen interpretedtocorrespondtoancestralmetazoanchromosomes,andcorrelationsbetweeninferredratesofgene movementbetweenALGsacrossthemetazoantreesuggestthattheseancestrallinkagerelationshipsareconserved throughtheactionofselectiveconstraintsonasubsetof genes[11]. Therelativelysmallnumberofgenomesfromanciently distinctmetazoanlineagesandthefragmentednatureof draftgenomeassembliesstilllimitboththesearchforancientwholegenomeduplicationsandthepowerofthe datatoconstrainmodelsofchromosome-scalegenome structureevolution.WhileWGSsequencingtechnology andassemblymethodsareactiveareasofresearchand technologicaldevelopment,andhaveimprovedatadramaticpaceinrecentyears,highquality denovo assembly oflarge,complexmetazoangenomesremainsadifficult andresource-intensiveproblem.Withoutgeneticorphysicalmaps,orrelianceonahigh-qualityreferencegenome ofaclosely-relatedspecies,WGSsequencingprojectsstill typicallyproduceassembliescontainingthousandsofscaffolds,hundredsofscaffoldsincorrectlyjoiningsequence fromdifferentchromosomes,orboth[12]. Next-generationsequencinghasgreatlyreducedthe costofconstructinghighdensitygeneticmapsbyeliminatingtheneedtodevelopandgenotypepolymorphic markersindividually[13].Thishasbeenachievedeither byfocusingsequencecoveragewithinoradjacenttogenomicregionsofdistinctbiochemicalcharacter,suchasrestrictionsiteswithrestrictionsiteassociatedDNA sequencing(RAD-seq)andrelatedmethods[14,15],orby combininginformationacrossregionsusingareference genomesequence[16,17].WhileRAD-seqisapplicableto organismslackingareferencegenomeassembly,itisnot directlyapplicabletocomparisonsofgenomeorganization acrosslongevolutionarytimespansbecausesuchcomparisonsrelyontheidentificationofhomologoussequence markers(typicallyprotein-codinggenes),whichtypically haveonlyasmalloverlapwiththerestriction-associated markers. Herewepresentagenotype-by-sequencingmethodfor constructingahigh-densitygeneticmapusinglowcoverage,low-cost,wholegenomesequencingdatafrom theoffspringofawildcross.Inthisjointassemblyand mapping(JAM)approach,thetraditionallyindependent andsequentialstepsofgenomeassembly,polymorphic markeridentificationandgeneticmapconstructionare combined.Existingassemblersexpectlowerdensitiesof sequencepolymorphism,deepercoverage,greatercomputermemoryormoreaggressivequalitytrimmingthat decreasesequencecoverage[18-20].Ourcurrentimplementationfocusesonconservativeassemblyofshort scaffoldssufficientformapconstruction,butourresults suggestthatfurtherintegrationofgeneticmappinginformationwithinwholegenomeshotgunassemblymethods canbeacosteffectivewaytoproduceassembliesoflarge, complexgenomeswithchromosome-scalecontiguity. Wehaveappliedthisapproachtoproduceageneticmap ofthegenomeoftheAtlantichorseshoecrab, Limulus polyphemus .Horseshoecrabsaremarinearthropodswith afossilrecordextendingback450millionyears[21].They exhibitremarkablemorphologicalstabilityovertheirlong evolutionaryhistory,retaininganumberofancestral arthropodtraits[22],andareoftencitedasexamplesof “ livingfossils ” L.polyphemus hasagenomeabout90%the sizeofthehumangenome.Itisanimportantspeciesfrom ecological,commercialandconservationperspectives[23], thathasbeenusedasamodelsystemforresearchin behavioralecology,physiologyanddevelopment[24].The mapandSNPmarkersdescribedherewillbearesource forthe L.polyphemus genomeproject,researchinhorseshoecrabpopulationbiology,andcomparisonsofmetazoangenomeorganization.B yanchoringproteincoding genestothismap,weareabletoextendanalysisofancestrallinkagegroupsandwholegenomeduplicationstothe cheliceratelineage.DatadescriptionApairofnaturallyspawninghorseshoecrabsandtheir eggswerecollectedfromtheirnaturalhabitatonthe beachatSeahorseKey.Thelarvaewerehatchedatthe lab4weekslaterfromthecollectingdate.Thetissue samplesfromthethirdwalkinglegsoftheparental horseshoecrabsand34larvaewereusedforDNA extraction,librarypreparationfollowingmanufactures ’ standardprotocols.Twoparentsand34larvaewereindividuallybarcodedduringlibrarypreparation.Illumina paired-endlibrarieswithinsertsizesofapproximate 300bpwerepreparedforeachsample.Theselibrarieswere pooledtogetherforthesubsequentsequncingontheIlluminaHiSeq2000platformatMedicalCollegeofWisconsin SequencingServiceCoreFacility.Atotalof1.7billion 100bpparied-end(PE)readswereobtainedafterthequalityfiltering.Thetotalsequencingcoveragewasestimated as38.9basedonthe k -merfrequencydistribution.The rawsequencingdatacanberetrivedfromNCBISRAvia NCBIBioProjectaccessionPRJNA187356.AnalysesAssemblyandmappingTheJAMmethodisdesignedtoproduceacombined assemblyofpolymorphicsequences,taggedbygenomic regionswithamaximumofonesinglenucleotidepolymorphisms(SNP)per k -merwindow(Methods).Startingwith genomicreadsfromamatingpairofadult L.polyphemus and34offspring(100bppaired-endreadson300bpNossa etal.GigaScience 2014, 3 :9 Page2of21 http://www.gigasciencejournal.com/content/3/1/9

PAGE 3

inserts),weanalyzed1.7billionreadscontainingatleast onehigh-quality23-mer. FittingPoissonmodelsforunduplicatedsequencetothe frequenciesoffiltered23-merssuggeststhat1.1billion genomiclociareuniqueatthisresolution(Figure1).This fitassumesthatthevastmajorityofmappablegenomic locihaveonlyoneortwoallelesrepresentedinthe parents ’ fourhaploidcontributions,whichgivesrisetothe fourcomponentsplottedinFigure1.Ofthesesufficiently uniqueloci,63%aremodeledashomozygous,27%as pairedmajor-minoralleles,andtheremaining10%astied allelepairs.ThecorrespondingSNPs,ifalwaysatleast23 basesapart,wouldbe1.6%ofbasesintheuniqueloci. Dividingthetotalnumberoffiltered23-mersbythe modeledhomozygousdepthofcoverage d =38.9yieldsan estimatedgenomesizeof2.74billionbases,consistent withthemeasuredDNAcontentof2.8pg(978Mb 1pg DNA)[25]. Wecategorizespecific23-mersbytheireditdistancesto others:havingnoneighborswithinasinglebasesubstitution(uniquetags)orwithasinglemutuallyuniqueonesubstitutionneighbor( “ SNPmerpair ” tags).Asubsetof these,includingSNPmerpairsforapproximately7.9millionSNPs,constitutethetagsusedforcontiggingand scaffolding.TheSNP-merpairsaccountforapproximately 45%ofthemodeledfractionofalleles,theothersmissed fromsimilaritytoothersequences(e.g.,duetorepeats)or distancefromeachother(becauseofindelsormultiple SNPsper23-mer). Chainingthese23-merstogether(seeMethods)produces aninitial6.6millioncontigs,3.9millionofwhichare linkablebypairedreadsforscaffolding.ApplyingBambus [26]produces944,000scaffoldsspanning1.3billionbases (Table1).Thesescaffoldsserveasmarkersincorporating multiple23-mertags,includingSNPmerpairsusedto identifyhaplotypes. Afterassembly,themeandensityofSNPsacrossthefour parentalhaplotypesinassembledregionswasestimated basedonreadre-alignmentstobe7.6perthousandbases. WejointlyinferredthephasesoftheseSNPsandsegregationpattern(offspringgenotypes)inthemappingcrossfor eachmarkerinamaximumlikelihoodframework (Methods).Wefocusedonthe91,320markerswithatleast 18inferredbi-allelicSNPsforconstructingthelinkage map.Thesemarkersgroupedinto1,908high-confidence mapbins(i.e.,uniquesegregationpatterns,assumedto correspondtolociinthegenomeuninterruptedbymeiotic recombinationinthecross[27].Mapbinsfellinto32 linkagegroups(Figure2),closetothe26pairs(2N=52) previouslyfoundinacytogeneticanalysisoftwochromosomespreads[28].Twentymapbinswereremovedfor havinginconsistentpositionsinthematernalandpaternal maps,and12weresingletons. Toestimatethefrequencyofincorrectgenotypecalls asafunctionoftheloglikelihooddifferencebetween thecalledandalternativegenotype(genotypeconfidencescore),includingcontributionsfromuncertainty inSNP-meridentification,assemblyandsamplingnoise, wecarriedoutasimulationofthelibrarypoolingand sequencing, k -merassemblyandgenotypeinferenceprotocols,usingthesequenced Cionaintestinalis genomeas astartingpoint. Figure1 FittingPoissondistributionstoLimulus23merfrequencies. Thedistributionofsequenced23-mers,modeledassamplinggenomic locithatarehomozygous(singleallelesharedbyallhaplotypes),havetwoallelesthataretied(AandapresentinparentsasAAaaorAaAa), orhavetwoallelesinamajor-minorrelationship(presentinparentsasAAAaorAaaa).Alleleswhoseparentalcontributionsarehomozygous, tied,majororminoreachhaveafrequencypeakcorrespondingtotheirdistinctivefractionoftheoveralldepthofgenomicsequencingd.Loci sharingsimpleorrepetitivesequencenotsufficientlyuniqueata23-merscalecontributetoalongtailofftherightedgeoftheplot. Nossa etal.GigaScience 2014, 3 :9 Page3of21 http://www.gigasciencejournal.com/content/3/1/9

PAGE 4

Inthesimulated C.intestinalis dataset(Methods),a singlestretchedexponentialdistributionprovidedagood fittothefrequencyofgenotypecallingerrorsasafunctionofthecallconfidencescoreforscoresupto6,or downtoerrorfrequencyofapproximately1%.Theobservederrorfrequencydeclinedmoreslowlyforhigher confidencescores.Theminimum x2fitusedforestimatingthegenotypingerrorrateinthe L.polyphemus mapbinswas pes a1eŠ s c 1 b 1 a2eŠ s c 2 b 2,withparameter values a1=0.49, b1=2.08, c1=1.26, a2=5.47, b2=0.17, c2=0.16(Figure3). Applyingthismodeltothe L.polyphemus marker genomecalls,weestimatedthatthegenotypecallingerror rateinthemapbinrepresentativemarkerswas0.0099. Weobservedthat51%ofadjacentmapbinpairsareseparatedbyasingleinferredrecombinationeventinthe cross,and94%areseparatedbythreeorfewerrecombinantsineachparent. Ofthe91,320markerswithatleast18putativeSNPs, 84,307(92%)wereassignedtotheirclosestmapbinswith athresholdof(Methods),foranestimatedgenome-wide averagedensityofonemappedsequencemarkerevery 32kb.Ameanof45markersweremappedtoeachmap bin,andthenumberofmarkersmappedwasusedtoestimatetherelativephysicalsizeofmapbins.Approximately 46%ofthescaffoldswith12 – 17SNPscouldbeplaced withthesamethreshold,foranadditional32,688markers, oronemarkerevery23kb. Thetotallengthofthescaffoldsassignedtomapbins was411Mb,andtheycontained2.67millionbi-allelic SNPsassignedaphasewithaposteriorprobabilityofat least0.99.Ofthese,72%wereinferredtobeuniquetoone ofthefourparentalchromosomes.Thisisclosetothe 74%predictedunderthefinitesitesneutralcoalescent modelgiventheobservedSNPdensity[29].SequencecompositionandrecombinationrateInthescaffoldslongerthan1kb(N=378,506andmean length=2.9kb),the G / C basecontentwas33.32.8%, andthelocalrelativefrequencyofCpGdinucleotides wasbimodallydistributed,withabout30%ofsequences exhibitingdepletionofCpG.TpGandCpAdinucleotideswereover-representedonaverageandtheirlocal densitiesnegativelycorrelated(r= Š 0.54,p<2.2e-16) withCpGdensity,suggestingongoinggerm-lineCpG methylationforafractionofthegenome[30]. Themeanmaternalandpaternalrecombinationrates wereestimatedtobe1.28and0.76centimorgansper megabaserespectively,consistentwithexpectationbased onthenegativecorrelationbetweenrecombination intensityandgenomesizeobservedinpreviousstudies [31].Wedidnotobserveevidenceofsegregationdistortionforanymapbins.Thecorrelationinlocalrecombinationratesintwoparentsacrossthegenomeis estimatedas r2=7.1%; p <1e-29,suggestingconsiderable variationinrecombinationlandscapebetweentwosexes inlimulus[32-34].Apositivecorrelationbetweenlocal recombinationrateandlocalSNPdensitywasobserved ( r2=9.7%; p <1e-40),whichisconsistentwithprevious observationsinhumanwithcomparablecorrelationcoefficient[35].AncestrallinkagegroupconservationWefoundthat34,942scaffoldshavesignificantsequenceconservationwith10,399predictedproteinsof thetick Ixodesscapularis :like Limulus ,achelicerate, butonewithawell-annotatedgenome[36].6,246of thesehitsformedreciprocalbestpairs,ofwhich5,775 (92%)couldbeplacedonthelinkagemapatathresholdof p <.Thesewereusedasconservedmarkersfor comparisonsofgenomeorganization.Whenlinkage groupsweredividedinto108non-overlappingbinsof 1,000markers,52hadsignificant( p <0.05,afterBonferronicorrectionfor1,944pairwisetests)enrichmentin sharedorthologs(or “ hit ” )withatleastoneofeighteen ancestralchordatelinkagegroups[7].AhiddenMarkov modelsegmentationalgorithm[6]identified40breakpointsinALGcompositioninthelinkagegroups. Approximately72%ofthegenomeisspannedby53 interveningsegmentsthathitoneor(foreightofthem) twoALGs(Figures4and5).EachoftheeighteenancestralALGshasatleastonehitamongthe45 segmentswithauniquehittotheALGs.WholegenomeduplicationsWholegenomeduplication(WGD),orpolyploidization isarare,butdramaticgeneticmutationeventwhich doublesthesizeofagenomeandcreatesaredundant pairofcopiesfromeverygene.Becauseitcreates Table1 K -mercontigandscaffoldstatisticsAssembledCountTotal(bp)Avg.span(bp)n50span(bp) k -mercontigs6,614,4341,240,275,515188418 Linkablecontigs3,925,8441,137,576,911290460 Initialscaffolds944,2461,261,263,17213363047 Referencescaffolds944,2461,295,334,51513722930 Referencebases1,131,458,74411982553 Nossa etal.GigaScience 2014, 3 :9 Page4of21 http://www.gigasciencejournal.com/content/3/1/9

PAGE 5

redundantcopiesofgenesforentirebiochemicalpathwaysandgeneticnetworks,ithasbeenproposedthatit createsuniquerawmaterialfortheevolutionofnovel biologicalfunctionsandincreasedcomplexity.HomeoboxgeneclustersHomeoboxgenesencodealargefamilyoftranscription factorsinvolvedindiverseembryonicpatterningandstructureformationprocessesofeukaryotes.Asaparticular Figure2 Eachofthenumberedblocksrepresentsoneofthe32linkagegroupsofthe L.polyphemus geneticmap,andiscomposedoffour columns:Twobandsofthetriangularmatricesinwhichthecolorscaleindicatesthefractionofsamplesshowingrecombinationbetween pairsofmarkers;maternalrecombinationfrequencyisshownontheleft,paternalontheright. Acolumnlabeled “ ALG ” indicatessegmentsof significant( p <0.05inFisher ’ sExactTest,afterBonferronicorrectionformultipletests)conservationofgenecontentwithancestralbilaterianlinage groups.ThecolumnlabeledHOXshowsthemappositionsandtypesofpredictedhomeoboxtranscriptionfactorgenes.Thetwocolorscalesarefor: recombinationfrequencybetweenpairsofmarkersandlogp-valueforenrichmentingenecontentwithancestrallinkagegroups. Nossa etal.GigaScience 2014, 3 :9 Page5of21 http://www.gigasciencejournal.com/content/3/1/9

PAGE 6

subfamilyofhomeoboxgenes,theHoxclusterisknownto controlmetazoanbodypatte rningalongtheanteriorposterioraxis.Weidentified155scaffoldswithsignificant homologytopredictedcheliceratehomeoboxgeneseque ncesinpublicdatabases.Weclassifiedthesesequencesinto homeoboxsubfamilies(Methods)andplacedthemonthe mapbybesthit.Twolargeclu stersofHoxgenesarefound onlinkagegroup(LG)15andLG21,eachcontainingmultiplemembersoftheanterior, centralandposteriorclasses. Therearealsotwoparahoxclusterhomologs,eachwith threehomeoboxgenes: gsx and cdx orthologsandathird homeoboxgenenotconfidentlyassignedtoasubfamilyin ouranalysis(LG5andLG19).Therearetwosmallerclusterscontainingmultiplehoxgenes(LG18andLG20),and clustersofotherhomeoboxgenes,includingmembersof the msx lbx nk evx and gbx families(Figure2).GenomicdistributionofparalogousgenesWGDcreatesmanypairsofduplicategenesor “ paralogs ” Thedistinctivefeaturesofthesegeneshavebeenusedto Figure3 Genotypecallerrorrateasafunctionofcallconfidencescoreforbinsof10,000callsinsimulated Cionaintestinalis genome data. Thestippledblueregionshows95%confidenceintervalsoftheBayesianposteriorprobabilitydistributionoftheunderlyingerrorrate computedfromtheBetadistribution Beta ( ne+1, nc ne+1)conjugatetheassumedbinomialdistributionofobservederrors,where neand ncare thenumberoferrorsandnumberofcallsineachbinrespectively.Theredcurveshowsthebestfiterrormodel a1e s c 1 b 1 a2e s c 2 b 2,withparameter values a1=0.49, b1=2.08, c1=1.26, a2=5.47, b2=0.17, c2=0.16. 2reduced=0.82. Figure4 Limulus-Humanmacro-syntenydotplot. BluepointsindicatethepositionofhumangenesinreconstructedancestralchordateALGs (verticaldisplacement)andtheircandidateorthologsinthe30 L.polyphemus linkagegroups(horizontaldisplacement). Nossa etal.GigaScience 2014, 3 :9 Page6of21 http://www.gigasciencejournal.com/content/3/1/9

PAGE 7

inferWGDeventsinfungal,vertebrateandplantgenomes [2-4].Weexaminedthegenomicdistributionof2,716 pairsofcandidateparalogousgenemarkersin L.polyphemus forsignaturesofWGD.In45%ofthesepairsboth markersmappedtothesamechromosome,compared with5.30.5%in1,000datasetswithrandomly-permuted paralogousgeneidentities.Themappedpositionsofpairs withinthechromosomeswerehighlycorrelated(average r2=0.81,andexceeding0.95for8ofthelargechromosomes;Figure6),suggestingthatmanyofthepairsrepresentrecenttandemgeneduplicatesorsinglegenes fragmentedacrossmultiplemarkers.Inthefollowing,these Figure5 Limulus-Humanmacro-syntenydotplotasinFigure4,showingbreaksintroducedbyhiddenMarkovmodelsegmentationof thelinkagegroupsasverticallines. Figure6 Genomicdistributionofcandidateparalogs. Themappositionsofpairsofputativelyparalogousproteincodinggeneswithinthe L.polyphemus genomeareplottedwithbluepoints.Pairsarebiasedtowardnearbymappositions,andthereforeconcentratedalongthe diagonal.Also,paralogssplitbetweenlinkagegroupsaresignificantlyclusteredinto “ paralogons ” Nossa etal.GigaScience 2014, 3 :9 Page7of21 http://www.gigasciencejournal.com/content/3/1/9

PAGE 8

same-chromosomeparalogsarereferredtoas “ tandem ” duplicates. Inter-chromosomalduplicat esareclusteredintoconservedparalogousmicro-syntenyblocks(or “ paralogons ” [3]):thereare25pairsofloci,eachwithatleastsix( mp =6) independentparalogpairsclusteredwithamaximum gap(max-gap)of300markersbetweenadjacentparalogsin eachcluster.Paralogpairsare consideredindependentif theyarebasedonhomologytoadistinctout-groupgene, toguardagainstrelyingoneithermultipleexonsofthe samegene,orrecently-duplicatedgenesasindependent evidenceofancientsegmentalparalogy.Theseclustersspan 25,044markers,or30%ofthemap,afterremovingredundancyfromparalogonswithoverlappingfootprints.In 1,000datasetswithrandomly-permutedparalogousgene identities,themaximumnumberofsuchclustersobserved was11;themeanandstandarddeviationwere3.91.0. Theobservedclusteringintoparalogonswasgreaterthan thatintherandomizeddatasetsoverabroadrangeof choicesofmax-gapandmp.Forexample,formax-gap= 100, mp =3thereare52clustersvs.3.51.9,range0 – 10; formax-gap=500, mp =9thereare12vs.2.91.7,range 0 – 9.Becauseofthelargeproportionofapparenttandem geneduplicates(45%),thisrandomizationschemeincreases thenumberofinter-chromosomalparalogpairsrelativeto thedata,makingitaconservativesignificancetestforinterchromosomalparalogcluste ring.Whengeneswithtandem duplicatesareexcludedfromtherandomizations,the observednumberofclustersisgreaterthanthemaximum observedin1000randomizationsforallthecombinations ofmax-gapintheset(100,200,300,400,500,600)and mp intheset(3,4,5,6,7,8,9).23max-gap=600, mp =7 clustersspan59%ofthemap,comparedtorespectivemean numberandmapcoverageof3.31.8,and136%inthese randomizations. Amongthemarkerpairsmappingtodifferentchromosomes,wefoundasignificantexcessofpairsrelatingsegmentsderivedfromthesameALGrelativeto randomizationcontrols(247pairsvs.10211, p <0.001 inrandomizationsofallgenes;202vs.467whengenes intandemduplicatesareexcluded).Thispatternisconsistentwiththecreationofthesesegmentsbyduplication (ratherthanfission). Themax-gapclustershaveasignificantamountofoverlapamongtheirfootprints.Forexample,thefootprintsof themax-gap=600, mp =7clustershadatotallengthof 72,072markers,butanetfootprintafterredundnacy removalof49,545markers.Agenomicregion a whichhas conservedsyntenywithtwootherregionsofthegenome ( b and c )couldariseeitherfrommixingofadjacent regions(throughlocalgenomerearragements)withhomologyto b and c respectively,orbysuccessiveduplications. Inthelattercase b and c arealsohomologous.Weexaminedtherelationshipsamongtheparalogonsforevidence ofsuccessiveroundsofduplication.Weconsidereda graphinwhichnodescorrespondtomerged,non-redun dantparalogonfootprintregions.Nodesareconnected withedgesifamax-gapclusterconnectsthetwonodes. Theaverageclusteringcoefficientofthisgraphisequalto theprobabilitythatfootprints a and c shareamax-gap cluster,giventhatthereareedges( a b )and( b c )inthe graph.Wecomparedtheclusteringcoefficientstothose foundinrandomErd s-Rnyigraphswiththesame numberofnodesandedgeprobabilityastheobserved graph.Wefoundthattheobserveddatashowssignificantlymoreclusteringthantheserandomgraphsfora widerangeofchoicesofmax-gapand mp.Forexample, formax-gap600, mp=7,theaverageclusteringcoefficientis0.19,while10,000randomgraphshadcoefficientsof0.0340.042, p =0.0039.AgedistributionofparalogousgenesBecauseWGDeventscreatemanyparalogsatthesame time,theyleavecharacteristicpeaksintheagedistribution ofparalogousgenes.In L.polyphemus thedistribution showspeakscenteredat0.71and1.34substitutionsper synonymoussite(Ks),valueswithintheapproximately linearresponserangeofKsestimatestoWGDage[37] (Figure7).Forcomparison,thesynonymoussitedivergencebetweenanAsianhorseshoecrabspecies Tachypleus tridentatus and L.polyphemus hasamodeof0.35.The commonancestorofthesespecieshasbeenestimatedto havelived114 – 154millionyearsago(MYA),coincident withtheopeningoftheAtlanticocean[38],suggestinga WGDevent230 – 310MYA,andpossiblyanolderone 450 – 600MYA.DiscussionOurresultsdemonstratethatalowcost,combinedapproachtowholegenomesequencingandgeneticmapping canbeusedtoefficientlycreateaveryhighdensitygenetic recombinationmapforanon-modelorganismwitha largegenome.Becausetheapproachusesgenome-wide sequencing,alargenumberofsequencemarkerscanbe anchoredtothemap,allowingcomparisonsofgenome organizationatthechromosomescaleoververylargeevolutionarydivergences.Theidentificationofchromosomal segmentswithsignificantgenecompositionhomologyto eachofthechordateALGsdemonstratesthatthepredominanceoffusionandmixingofancestrallinkagegroups previouslyobservedinanalyzedecdysozoangenomes[10] isnotancestralto,oruniversalintheclade. Themapallowsquantitativecharacterizationofother featuresofchromosome-scaleorganization,suchasthe correlationbetweenlocalrecombinationrateandpolymorphismlevels.Similarpositivecorrelationsbetween localrecombinationrateandpolymorphismlevelhave beenobservedinothermetazoansincludinghumansNossa etal.GigaScience 2014, 3 :9 Page8of21 http://www.gigasciencejournal.com/content/3/1/9

PAGE 9

[39-41]andplants[42,43].Futurecomparisonswithmore closelyrelatedchelicerateswillallowteststodistinguish whethertheseratesarepositivelycorrelatedwithinterspecificdivergence,consistentwithaneutralprocessof correlatedmutationandrecombinationrates[35].Alternatively,theassociationcouldbeexplainedbyhitchhiking andbackgroundselection[44]. Theenrichmentofinter-chromosomalparalogpairsin segmentsofthesameALGoriginisconsistentwiththeir creationbyduplication(rath erthanfission),although becausesmall-scaleduplicationisbiasedtowardlocal (tandem)duplication,fissio nofsegmentscouldalsoleave behindanenrichmentofparalogs.Suchamechanism, however,wouldnotcreatetheobservedorganizationof paralogs,thatis,theirclusteringinto “ paralogons ” .Thefact thattheseparalogonsspanalargeportionofthemap(59%) suggeststhatitwasawholeg enomeduplication,rather thansegmentalduplicationsthatgaverisetothepattern. Theexistenceofduplicatedhoxandparahoxclusters onfourdifferentchromosomesishighlysuggestiveof multiplewholegenomeduplication.Hoxclustershave notbeenfoundinduplicatecopiesexceptinvertebrates wheretheyhavebeencreatedbywholegenomeduplication,andhaveonlyrarelybeensubsequentlylost. Thedouble-peakedshapeofthedistributionofsynonymoussitedivergencebetweenpairsofparalogs,combinedwiththeexistenceoftwosmallclustersofHOX genesinadditiontothetwocompleteHOXclusters suggeststhattheremayhavebeentworoundsofwhole genomeduplicationinthehorseshoecrablineage. WGDsprecededmajorspeciesradiationsinvertebrates,angiospermsandteleostfish,andtheimportance oftheirroleinevolutionisthesubjectoflong-running debate[1-4].Thediscoveryofwholegenomeduplication inaninvertebrate,andduringhorseshoecrabs ’ longand famouslyconservativeevolutionaryhistorysuggeststhat sucheventsmayhavebeenmorecommonthanpreviouslyassumedinmetazoanevolution,andthatwhile theymayhaveprovidedrawmaterialforadaptiveevolutioninsomecases,theyarenotevolutionarydrivers.MethodsJointassemblyandmapping(JAM)overviewBarcodedgenomicDNAlibrarieswerecreated,pooled, andsequencedinfourlanesontheIlluminaHiSeq2000 platformforamatingpairof L.polyphemus and34 offspring. Figure7 Distributionofestimatednon-synonymous(Ka;top)andsynonymous(Ks;bottom)andsequencedivergenceratesforpairsof putative L.polyphemus paralogs(left)and L.polyphemus T.tridentatus orthologs(right). Nossa etal.GigaScience 2014, 3 :9 Page9of21 http://www.gigasciencejournal.com/content/3/1/9

PAGE 10

TheJAMmethodproceedsthroughthreemajorphases: 1.ThefrequenciesofDNAsub-sequencesoffixedlength k ( k -mers)areprofiledtocharacterizethequality,uniqueness, polymorphismsandrepetitioningenomicreads,usingsoftwarewedevelopedbuildingonworkfromtheAtlasassembler[45].Allelicpairsof k -mersrepresentingalternate formsofSNPsareidentified andtrackedthroughthesubsequentsteps.2.Contigsareassembledonagraphof unique k -mersandpairedSNP k -merssampledtoreduce memoryusage,thenorderedandorientedusingthe Bambusscaffolder[26,46].Ea chmulti-SNPscaffoldistreatedasasinglemarkerforthelinkagemappingsteps3.The pairedSNP k -mers(ineachscaffoldarecombinedwiththe read,mate-pair,andparent-oroffspring-libraryassociationsoftheirallelesforhaplot ypephasingandconstruction ofahighdensitygeneticlinkagemap.ThesoftwareispublicavailableasopensourcesoftwareatGitHub[47].SamplingandsequencingTheparentalhorseshoecrabsandtheireggswerecollectedfromtheirnaturalhabitatonthebeachatSeahorse Key,anislandalongthewestcoastofnorthFlorida,on27 March2010.Thisnaturallyspawningpairwereobserved astheeggswerebeinglaidandfertilized(fertilizationis externalinthisspecies,i.e.,theeggsarefertilizedinthe sandunderthefemaleastheeggsarebeinglaid).The tissuesamplewerecollectedfromthethirdwalkinglegs ofthisparentalpair.Wemarkedwheretheeggsamples werelaidandreturnedafewhourslateranddugupthe nest,thenremovedthefertilizedeggs.Wealsohave conductedpaternityanalysesthatshowthatfertilizationis bytheassociatedmaleandnotbyextraneousspermthat mightbeatthenestingsite(inthiscasethedensityof nestingfemaleswaslowonthisdaysoweknowthatthe eggswecollectedwerefromthepairweobserved). Trilobitelarvaewererearedinplasticdishesaspreviously describedandhatchedfromtheeggs4weekslater[48]. TissuesamplesandlarvaewerepreservedinRNALater. GenomicDNApurificationandlibraryconstructionwere carriedoutusingQiagenDNAEasy,IlluminaTruSeqand Nexterakits,followingmanufacturers ’ protocols.Barcoded sampleswerepooledandsequencedontheIllumina HiSeq2000platform. Limulus larvaewereprocessedasfollows;eachlarva, suspendedin100 L ofRNAlaterandstoredat Š 80C ina1.5mLEppendorftube,wasthawedonice,after whichRNAlaterwasremoved.DNAwasextractedusing theQiagenDNAEasykitpermanufacturer ’ sprotocols. DNAwasquantifiedusingpicogreenDNAquantitation kit.ToprepareTruSeqlibraries,DNAwasfirstpurified anothertimeusingzymogenomicDNAcleancolumns permanufacturer ’ sprotocols.Adult L.polyphemus DNA waspreparedasabove,butusingclawtissueratherthan wholelarvae.AllDNAextractsweretestedbygel electrophoresistoensureDNAwasnotdegraded.TruSeq librarieswerepreparedatUniversityofGeorgia ’ sGeorgia GenomicsFacility.1 – 5 gofsampleDNAwassubjected tofragmentationusingCovarissonicator.Fragmented DNAwasthenusedforlibraryconstructionusing IlluminaTruSeqlibraryprepkits.Librarieswerepooled togetherinequimolaramounts(for10larvae)andused forthefirstsequencingruninseparatelanesforthe parentalandlarvalpools.Forlarvalsamples11 – 34,library prepwasswitchedfromTruSeqtoNexterakits.Nextera librarypreparationwasperformedaccordingtomanufacturer ’ sprotocol.TheNexteralibraryproductwasquantifiedbypicogreen,andfragmentsizedistributionwas checkedbyusingLonzaflashgel,toensurethatfragment sizedistributionwasbetween300 – 1,000bp.Sample librarieswerepooledinequimolarconcentrationsand sentforthesecondsequencingrunintwolanes,eachon apoolof12larvae.Bothsequencingruns,comprisingfour librarypools,wereperformedontheIlluminaHiSeq2000 platformatMedicalCollegeofWisconsinSequencing ServiceCoreFacility. Atotalof1.7billionIlluminareadsqualifiedfor kmer analysisandassemblybycontainingatleast23consecutiveq20bases.Thematernallibraryaccountedfor13%of thesereads,thepaternallibrary7.4%,andthe34offspring librariesfor2.4%onaverage,0.64%atminimum.k-merdecompositionWedeterminedalowerboundonthe k -mersizelong enoughforagivenexpectationofuniquenesinarandom genome.Whileincreasing k reducedtherateofcoincidentallyrepeated k -mers,italsoreducedtheeffectivedepthof coverageduetountrimmederrorsandedgeeffectsat readends — andincreasedthecasesofmultipleSNPs per k -merlocus,whicharenottrackedinourcurrent softwareimplementation.Wecanapproximatelymodel agenome-scalestring G ofrandomnucleotidesas G samplestakenwithuniformprobabilityfromthespace ofall k -mers(ofsize4k/2forodd k -merstreating reverse-complementsassame;slightlymoreforeven k ).ThePoissondistributionthengivestheprobability thatalocationin G hasitsown,unique k -mer(shared withnootherlocation)as eŠ ; where G 4k2 2 G 4kTheprobabilityofalocationsharingits kmer isthen; thus,tolimitthemaximumrate R of G -locationssharing k -mers,werequire k [log4( Š 2 G /ln(1 Š R ))].Forexample, foramammalian-scalegenomeofapproximately3billion bases,and R =0.1%,wechose k 22.For Limuluspolyphemus ,theAnimalGenomeSizeDatabase[25]reportsan estimatedhaploidgenomesizeof2.80pgand,aseachNossa etal.GigaScience 2014, 3 :9 Page10of21 http://www.gigasciencejournal.com/content/3/1/9

PAGE 11

picogramrepresentsalmostabillionnucleotidebase-pairs ofDNA,themammalian-scalechoiceof k applies[19,49]. Thislowerboundignoreschemicalandbiological sequencebiases,soselecting k forarealgenomeproject requiresattentiontoerrorrates,repeatstandemand interspersed,andgenomesize,allknownvaguely,ifat all,beforesequencing.Studyingthe k -merdistributions aftersequencingcanclarifythesegenomicpropertiesas weselect k tomaximizethenetyieldofcandidate k -mer tags,betweenerrorsandwithatmostoneSNPlocation, insequencingreads.WeconvertedIllumina/Solexa FASTQformat(payingattentiontothedifferentquality encodingsofthesoftwareversions)intoFASTAformat [50],masking(replacingwith ‘ N ’ )anybasewithPhredscale[51]qualitybelow20,andsoft-masking(representing inlowercase)otherbaseswithqualitybelow30.Forinitial trimmingexperiments,wevariedthesequalitythresholds asindicatedbelow.Westored k -mersinhashtableswith openaddressing[52],supportingodd k <=31.Wetallied foreach k -merabitvectorforpresenceorabsenceinup to64samplelibraries(36for Limulus parentsandoffspring),andanoverallcountofoccurrencesinalllibraries (countlimitedto64 Š 2 k bits).Wherethe kmerhash wouldbetoolargeforavailablememories,wesampledthe k -mersusingahash-slicingfactor S (mustbeprime). Representingeach k -merasanintegerin,slice s consistsofthose kmerswhoseremainderondivisionby S is s .Wecantabulateonesliceforarepresentativesample of1/ Skmers(forinitialestimationofdepthofcoverage andgenomesize)or,using S independentjobs,tocollect informationfor k -mersinallslices.Ourhashtablesstored odd-length k -merssothatreverse-complementary sequencescanbecombinedwithouttheambiguousorientationofpalindromicsequences(e.g.,ATCGAT). Afterselecting k asdescribedaboveandmakingafull tabulationof k -mercountsandbitvectors,wefiltered out k -mersnotexpectedtorepresentgenomicsequence. k -merswererequiredtohavethreecopiesinthetotal sequenceset,withatleastonecopyintheinitialrun andoneinthesecondrun.Thiswaspartlytofilterout incompleteadaptersequences,whichcanbedifficultto trim,butweredifferentinthetworuns. ExtendingmethodsdevelopedfortheAtlasassembler [45]toheterozygoussequences,Figure1givesarough decompositionofthe k -merfrequencydistributionfor 23-merswithquality 20,minimizingthesquareofthe residualsof k -mercountsonfrequencies3through70whilenotexceedingtheobservedcounts.Fourlinked distributionsmodelfractionsofthegenomeasmonoallelicorbiallelic:homozygousregionswith d =38.9-fold coverage(darkblue),minorallelescoveredat d /4 (green),tiedallelesat d /2(red)andmajorallelesat3 d /4 (purple).Thisfitisrobustenoughtoconfirmtheabundanceofmajor-minorallelepairs(27%of k -loci,vs.10% fortiedalleles),withthebroaderpeaksinthedatathan inthefittedcurvesconsistentwithlessuniform sampling(forexample,varyingcoverageofparentsand offspring).ThePoissondecompositionsuggestsadensity ofpolymorphismsof1.2%inmajor-minorallelepairs, basedondividingthemodelednumberofsuchsequencedpairsby k (assumingmostpolymorphismsare SNPsspacedatleast k basesapart),by d (theestimated depthofsequencing)andbytheestimatedgenomesize of2.74billionbases.SNPmeridentificationThefilteredkmercounts,computedinparallel,areloaded intoahashtablewithadditionalfieldstotrackkmersthat areuniquelywithinonemismatchofeachother.Because thisstepanalyzesall(non-error) k -mersinonetable,this requiresasinglelarge-memoryprocessor(ontheorderof 32GiB). Foreach k -mer,wecheckallits3 k one-substitution neighbors.The k -mersarepartitionedeachintooneof threecategories: unique :havingnoedit-neighborswithin onesubstitution; ambiguous :havingeithermultipleonesubstitutionneighbors,oroneneighborthathasmultipleneighbors;or partnered :uniquelypairablewith exactlyoneother k -merdifferingbyonesubstitution, such k -mersalsoknownfromnowonasSNPmersor SNPmerpairs.ForeachSNPmer,wesavethepositionof thesubstitution,abitmaskforthechange(transition, complement,ornon-complementtransversion),and whetherthecanonicalformofthepartnerinthetable hasthesamesenseorisreversecomplementedwith respecttothis k -mer. Onlypartneredandunique k -merswillbefurther tracked.Whilethislimitedmethodcannotidentify k -mers forgenomicSNPandnon-SNPlocationswithcomplete confidence,falsepairingormissedpairingshouldhave limitedeffects,asconfirmedbyassemblyexperiments withsimulated Ciona sequence(see Errormodelcalibration inMethods). Falsepairing ,duetocoincidentalsimilarities orrepeats,wouldcombinenodesofthe k -mergraph (seebelow)andcausenoiseinthescaffolding,haplotype phasing,andlinkageanalysis.Suchmisleadinglinksare minimizedbythe robustedge requirementsincontiggingandscaffolding,describedbelow. Missedpairing canhappenfromindelpolymorphisms,SNPsseparated byfewerthan k Š 1positions,failuretosequenceminor alleles,orambiguityduetotoomanysimilar k -mers. Ambiguouslynon-unique k -merswillbeskippedover (reducingconnectivityofthe k -mergraphifthereare toomanyinarow).Whereallelic k -mersmisidentified asuniquecauseconflictingedgesinthe k -mergraph, nodesforunpartneredmajoralleleswilleitherbechained intocontigswithflankinguniquesequenceorleftas orphanedfragments,andunpartneredminoralleleswillNossa etal.GigaScience 2014, 3 :9 Page11of21 http://www.gigasciencejournal.com/content/3/1/9

PAGE 12

beleftasorphanedfragments.Overall,errorsinidentifyingparneredandunique kmersshouldshortencontigs andscaffoldsandhidelinkage,notpromotefalselinkages. Table2showsthetotalsandpercentagesofthedifferentkmercategories,countingeachSNPmerpairasone k -mer.SNPmerpairsaccountfor16.3%oftheputative genomicallyunique23mermarkers;dividingby23gives usthefractionofbasesinthosemarkersthatareputativeSNPs:0.71%.Nodek-merselectionToreducethememoryrequirementsofour k -mer assemblygraph,weselectaapproximateone-tenthsubsetoftheSNPmerandunique k -mertags. InthecaseofatrueSNPatleast k1basesfromother SNPsandfromgapsinerror-freecoverageofeitherallele, therewillbe k coveringSNPmerpairs(providedthatcovering k -mersarealsouniquelypairable).Bytakingonly SNPmerpairswiththesubstitutioninparticularpositions, wecanreducethesizeofthegraphanditsredundancy. Analyzingthedistributionofsubstitutedpositionforall theSNPmerpairs,weobserveanenrichmentforsubstitutionsneartheends,probablyduetoproximityto low-qualitysequence.Byselectingpositions3,12and 21of23-baseSNPmers,weavoidthemostproblematic positionsandreducethisportionof k -mernodesbya factorof7.67. UnlikeforSNPmers,therearenocanonicalpositions thatidentifytheunique,unpaired k -mers.Severalmechanismshavebeenproposedforsampling k -mersinarepresentativeway[53,54].Weusethemorepseudo-random hash-slicingrule,alreadydiscussedabove,tosamplea singlesliceof k -mers:thosewhoseintegerencodingsare congruenttoaparticularslicenumber s ,modulo S (the hashslicingfactor).Wehavefoundthatonthefinished humangenome(resultsnotshown),hashslicingiseffectivelyaPoissonsampling,withsampled k -mersspaced accordingtoanexponentialdistribution. Acaveatinapplyinghashslicingisthattakingthe remaindermoduloaprimeisnotverypseudo-random forMersenneprimes(equalto2p Š 1forsome p ),when k -mersarerepresentedinbase-4encoding[52].We thereforepickaslicingfactorof11,thesmallestnonMersenneprimegreaterthanourSNPmersampling factor. Theresulting k -mersubsethas86.0millionuniqueunpaired k -mersand24.0millionSNPmerpairs,each reducedaspredicted,foratotalfactorof9.8reduction in k -mernodesforthenextstep.ContiggingandscaffoldingEach23mertag(unique k -merorSNPmerpair)inthe abovesubsetisanodeinthe k -mergraph.Nodesareconnectedwhenthecorresponding k -mersappearconsecutively inatleastonereadoftheinput(anyintervening k -mers havingbeenskippedduetosamplingorambiguity).The relativeorientation,distan ceandnumberofsupporting readsofthe k -mersisstoredintheedge.Whenconflicting distanceorrelativeorientationisobservedamongdifferent readsforthesamepairof k -mernodes,alledgesfromboth nodesinthecorrespondingdirectionareignoredin contigging. Thenodesofthe kmergraphrepresentDNAtagsand havedistinctupstreamanddownstreamends.Oneedge ateachendofanodeisidentifiedas robust ifsupported byasupermajorityofthereadsforalledgesinthatdirection:thenumberofsupportingreadsisgreaterthanor equaltoboth(1)twoplusthesumofthereadcountsfor allotheredgesinthatdirectionand(2)twicetheread countofthenext-mostsupportededgeinthesamedirection.Bythisconstruction,anodehasatmostonerobust edgeoneachend. Amutuallyrobustedgeisdefinedasonethatisrobust goinginbothdirectionsbetweenthetwonodesit connects. Contigsaretheconnectedcomponentsofthesubgraph consistingofmutuallyrobustedges.Singletonandcircular contigsarereportedfordiagnosticpurposes,butignored insubsequentanalysis.Eachretained “ k -mercontig ” of Table1thereforerepresentsachainofnodesforSNPmers andunique k -mersnotsharedwithothercontigs. Afterassemblyof k -mercontigs,weconnectthemin longerstructuresusingtheBambusscaffolder[26].Becausethecontigsdonotcontaindetailedreadinformation,wemaptemplates(readpairs)tocontigsbasedon shared k -mercontent,anddividingtheresultinggraph ofcontigslinkedbytemplatesintobatchessmallenough forBambustoprocess.Batchesaredividedsothatno contigsindifferentbatchessharenotemplates. WepresenttemplateslinkingcontigstoBambususing AMOSformat[46]forthereads(templateends)mapped toeachcontig.Readsareincludedonlyforthecontig withwhichitsharesthemost k -mers,ifthespanof those k -mersis k andtheotherread-endofthetemplate similarlyqualifiesinadifferentcontig.Bambusinferslinks betweencontigsbymatchingtemplateidentifiersshared byreadsindifferent “ linkablecontigs ” ,thenproduces scaffoldsaschainsofcontigsthatarelinkedwithconsistentorderandorientation. Table2K-mercategories,countingaSNPmerpairas one k -merk -mertype#DistinctPercentage Nopartner/unique946,431,90155.48% Partnered/SNPmer184,756,14910.83% Ambiguous574,557,29633.68% TOTAL1,705,745,346 Nossa etal.GigaScience 2014, 3 :9 Page12of21 http://www.gigasciencejournal.com/content/3/1/9

PAGE 13

Contiguousconsensusrepresentationsfor k -mercontigs andscaffoldsweregeneratedintwophases.Inthefirst phase,sequencespannedbyselectedSNPmersandsubset k -mers(seesectionsabove)arejoinedtogether,separated byanumberofNscorrespondingtothenumberofbases notspannedby k -mersinthesubset.Inthesecondphase, asinglepassismadethroughthereaddataset,and stretchesofNsthatarespannedbysinglereadsare replacedbythesequenceoftheread.SNPphaseandgenotypeinferenceEachscaffoldofthe k -merassemblyconstitutesacandidatemarkerformapping.Whilethedepthofsequence coverageoneachmemberofthemappingpanelistoo low(~1X)todirectlyinferthegenotypeofindividual membersofthemappingpannelatindividualSNPs,the tightlinkagebetweenSNPswithinmarkersmeansthat learningasample ’ sgenotypeatanyonerevealsitatthe others,effectivelyamplifyingthesequencecoveragebya factorproportionaltothenumberofSNPswithinthe marker.Thisisthesameprincipleexploitedingenotype bysequencing(GBS)approachestogeneticmappingin thepresenceofreferencegenomes,forexampleinrecombinantinbredlinesofreferencericestrains[16],andin crossesbetween Drosophila specieswithsequenced genomesHerewegenotypeoffspringinthecontextofa crossbetweentwooutbredindividuals,simultaneouslyinferringthephasesoftheSNPs(i.e.,whichbasesappears oneachofthefourparentalchromosomesinthecross). Whilethedatawillbeinsufficienttoinfergenotypesat manymarkers,allthosewhereconfidentinferencescan bemadecanbeusedtobuildthelinkagemap. Forthepurposesofgenotypeinference,amarkeris treatedasacollectionof m SNPs(indexedinthefollowing by i {1,2, … m }),thathavebeeninferredtobeclosely linkedonthegenomeviathe k -merassemblystep.Ifthe fourparentalchromosomesarelabeled a b inoneparent and c d intheother,thenthegenotypingproblemisto inferwhichofthefourpossiblesegregationstatesorgenotypes ac ad bc bd describeseachsampleateachmarker locus.Weindexsampleswith j ,anddenoteasample genotypeby gj. Weassumethatmarkersareverysmallcomparedtoa chromosome,andignorethepossibilityofarecombinationeventwithinindividualmarkers.Thedatausedfor inferenceoftheoffspringgenotypesconsistofthenumberofreadsfromeachbarcodedsample j showingeach ofthefourpossibleDNAbases b ateachvariableSNP position i ,whichwedenote nb ij. Ifthephase iofSNP i wereknown, i.e. whichbaseis presentineachofthefourparentalchromosomes,then achoiceofgenotype gjimpliesaspecifichomozygousor heterozygousstate sij S ={ AA CC TT GG AC AT AG CT CG TG }forSNP i insample j .Foragivenphaseand genotype,thelikelihoodfunctionforagivenSNPpositioninagivensampleisgivenbyeitherabinomial(for homozygousstates)ortrinomialprobabilitymassfunctionofthereadcounts,base-callingerrorrate ,andthe sitegenotype sij:L i; gj p nb ij i; gj Pm nb ij sij; n m m 1 n m ; ifsijhomozygous n! k!l!m! 2 3 m 1 2 3 2 0 B B @ 1 C C Ak l; ifsijheterozygous 8 > > > > > > > < > > > > > > > :where n isthetotalnumberofreadsatSNP i ; m isthe numberofobservationsofbasesnotin ij(i.e.,mismatches); k and l arethecountsforeachofthetwobases of ijforheterozygoussites.LikelihoodmaximizationSearchingforanoptimalchoiceofSNPphases iand samplegenotypes gjismadedifficultbytheexponential sizeofthesearchspace:forsegregatingbi-allelicSNP sitesthereare14possiblephasestoconsiderateach SNPsite,soforamappingpanelofonly20siblingsand amarkercontainingonly10SNPs,therearecombinationstoconsider.Insimulationtests,wefoundthata variantofexpectationmaximization(EM),aniterative likelihoodmaximizationmethodcanaccuratelyinfera largeproportionofmarkergenotypes. Toinitializetheiteration,theparentalsamplesanda randomlyselectedoffspringare,withoutlossofgenerality,assignedgenotypes( a b ),( c d )and( a c ).Ateach step,wecalculatetheconditionalprobabilitydistributionsoverthepossibleSNPphases p ( i)giventhe genotypeassignmentsaccordingto:pt i p ij gt Y Spnb i j i; gt = XkY Spnb k j k; gt !wherewehavelabeledthechosenvaluesforthegenotypesatiteration t collectivelyby g( t )g( t )nb i isthecombinedtotalnumberofobservationsofbase b at polymorphicSNP i forallsamplesincludedatiteration step t whichhavegenotype s ( i, gj)= Oneachiterationuntilallsampleshavebeenincluded,arandomlyselectedsampleisaddedtotheset aftercalculating p( t )( i).ThenthenextsetofgenotypeNossa etal.GigaScience 2014, 3 :9 Page13of21 http://www.gigasciencejournal.com/content/3/1/9

PAGE 14

assignments g( t +1)aredeterminedbychoosingthose thatmaximizetheexpectedvalueoftheloglikelihood: E j n ; gj log L gj; n ; Xip i log pnb ijj gj; iThesestepsarerepeateduntilgenotypesarebeingselectedforallsamples,andtheexpectedloglikelihood stopsincreasing.Attheendoftheiteration,the likelihood-maximizinggenotypesarereported,alongwith theloglikelihooddifferencebetweenthebestandsecond bestchoiceofgenotypeforeachsample,whichprovides anindicatoroftheconfidenceingenotypecall.Togauge convergence,thisprocedureisrepeated5timesforeach marker,withdifferentrandomchoicesofinitialconditions.MarkerswhichdonotidentifythesameMLgenotypemultipletimesinindependentrunsarenotincluded amongthehighconfidencegenotypecalls.MapbinsUniquemarkersegregationpatternswereincludedin thesetofmapbinsiftheymetoneoftwocriteria:(1)at leastthreeindependentmarkerswereinferredtohave thepatternindependently,or(2)thepatternwasinferredfromatleastonemarkerwithatleast20SNPs suchthatthemeanoftheestimatedprobabilitiesofthe inferredSNPphaseswasgreaterthan0.9.ErrormodelcalibrationThesequenceof14 Cionaintestinalis autosomeswere downloadedfromEnsembl[55].These14chromosomes wereusedasthetemplateinourgenomesimulation. Basedontheirsequencelength,Weusedamarkoviancoalescentsimulatormacs[56]togeneratefourhaploid samplesdrawnfromapopulationunderneutralWrightFishermodelwithpopulationmutationrateof0.012and populationrecombinationrateof0.0085.Usingthe C. intestinalis genomeasthereferencesequence,twodiploid parentalgenomeswereconstructedbasedonthemacs outputwithrealisticSNPandIndelmodelsinferredby severalpreviousstudiesonthe Ciona genome[57-59].We wroteaperlscripttosimulatethegenomesofoffspring generatedbythecrossofthetwosimulatedparents.The softwarepackagedwgsim[60]wasusedtogenerateIlluminapaired-endreadsbasedonoursimulatedgenomes ofbothparentsandoffsprings,withthecoverageof20X and5Xrespectively. Toestimatethefrequencyofincorrectgenotypecallsasa functionoftheloglikelihooddifferencebetweenthecalled andalternativegenotype,includingcontributionsfromuncertaintyinSNP-meridentification,assembly,andsampling noise,wecarriedoutasimulationofthe k -merassembly andgenotypeinferenceprotoc olsAmonghigh-confidence genotypecalls,theobservederrorfrequencywasafunction ofcallconfidencescorewaswell-fitbyasumoftwo stretchedexponentialfunctions,allowingassignmentof errorprobabilitiestoindividualgenotypecalls.LinkagegroupconstructionWeusethelinkage p -value pabbetweenpairsofmap bins a and b definedastheminimumoverthefourpossiblerelabelings r ofthematernalandpaternalchromosomesoftheBinomial p -valueforthenumberof matchinggenotypes: pab minr1 Š XmrŠ 1 in i 1 2n"# where n isthetotalnumberofsamplegenotypecalls (68inthepresentcase,or34ineachparent)and mris thenumberofmatchinggenotypesunderrelabeling r Weidentifiedmapbinswithsegregationpatternsindicatingeitherinconsistentplacementinthematernaland paternalmapsorgenotypingerrorwithadoublethresholdprocedureasfollows: 1.Mapbinswerepartitionedintolinkagegroupsby singlelinkageclusteringatathresholdof. pab< p1. 2.Withineachpartition,mapbinswhichformed articulationpoints(i.e.,nodeswhich,ifremoved, wouldcausethelinkagegrouptofallapartintotwo disconnectedsubgraphs;)inthegraphof, pab< p2, where p2> p1. Thisprocedureidentifiesmapbinswhichaloneaccount forthemergingofwhatwouldotherwisebetwodistinct partitions.Weusedthefollowingpairsofthresholds p1, p2toidentifyatotalof20mapbinsforexclusionfrom themap:10,10;10,10;10,10.Theremainingmarkersform locallyconsistentlinkagegroupsinwhichalllinkagesdefinedatthreshold p1arecorroboratedbymultiplelinkages at p2,fortheabovevaluesof p1and p2.MarkerorderingMarkerswereorderedwithineachlinkagegroupusing thefollowingprotocol.Withineachlinkagegroupaconsistentlabelingofthefourparentalchromosomeswas achievedbyconstructingagraph G inwhichnodescorrespondtomapbinsandedgesareweightedbylinkage p -value pab(asdefinedabove).Thelocalchromosome labelsareupdatedateachmapbinasitisreachedina traversaloftheminimumspanningtreeof G tothelabeling r thatmaximizes pabalongtheincidentof G used inthetraversal.Markerswithineachlinkagegroupwere clusteredbyhierarchicalclustering(marker-markerdistancemetric:cosineoftheanglebetweenthevectorsof recombinationdistancestotheothermapbins;distance updatingmethod:averagelinkage)intoabinarytreedataNossa etal.GigaScience 2014, 3 :9 Page14of21 http://www.gigasciencejournal.com/content/3/1/9

PAGE 15

Figure8 Unrootedphylogenetictreeofhomeoboxsequences(part1). NodesarelabeledwithBayesianposteriorprobabilities.Highly supportedpartitionsusedtoclassify L.polyphemus sequencesaredrawninred,withtheabbreviationfortheclassshowninlargeletters. L. polyphemus homeoboxsequencesnotgroupedintooneofthesehighlysupportedpartitionsareassignedtoclass “ ? ” .Foreaseofdisplay,a largesubtreeconsistingofHOXandparahoxgeneshasbeenprunedatthepositionlabeled “ HOX ” ,andisshowninFigure7. Nossa etal.GigaScience 2014, 3 :9 Page15of21 http://www.gigasciencejournal.com/content/3/1/9

PAGE 16

Figure9 (Seelegendonnextpage.) Nossa etal.GigaScience 2014, 3 :9 Page16of21 http://www.gigasciencejournal.com/content/3/1/9

PAGE 17

structurewithleavesrepresentingmapbins.Anodein therightsubtreeoftherootnodewasrotated,interchangingitsleftandrightsubtreesifitsleftsubtreewas notalreadycloser(inaveragerecombinationdistance)to themarkersoftheleftsubtreeoftheglobalroot;and similarlyfornodesintheleftsubtreeoftheroot.Aninordertraversalofthetreegeneratesanorderingofthe markers.Finallythreereversalsoftheorderofmarkers insegmentsofthemapwereaddedbasedonvisual inspectionoftherecombinationdistancematrix.Inthe finalmarkerordering,51%ofadjacentmapbinpairsare separatedbyasinglerecombinationeventinthecross, and94%areseparatedbythreeorfewerrecombinants ineachparent.PlacementofmarkersonthemapToanchoradditionalmarkerstothemap,wecomputed the pab(seeabove)betweenmarker a tobeplacedon themapandeachmapbin b .Marker a isanchoredto themapatthepositionofthebin b whichminimizes pabif pab<10Š 6.SNPdensityestimationIlluminareadsweremappedtotheassembledscaffoldsequenceswithstampy[61]usingdefaultsettings.Fora sampleof9,228scaffoldswithlengthsrangingfrom5.05.5kb,sequencevariantswerecalledwithSAMtools[62] usingavariantqualityscorethresholdof50,andignoring indelpositions. ASNPdensityof0.76%infourhaplotypescorresponds toapredictedrateofpairwisesequencedifferencesper siteof =0.0042underthefinitesitesmodelofmutation andtheneutralcoalescentmodeloftherelationships amongsampledalleles[63].EstimationoflocalrecombinationrateToestimatethelocalrecombinationrateforeachmap bin,wecomputedthelinearregressionofmapdistancein numberofmarkersonphysicaldistanceusingupto10 neighboringmapbinsineachdirectionalongthemap(or fewerforbinswithin10mapbinsoftheendofthelinkage group).Mapdistancewascalculatedfromrecombination fractionusingHaldane ’ smapdistance Š 1 2log1 Š 2 r [64].AncestrallinkagegroupconservationTocomparethegenomeorganizationin L.polyphemus to theancestralmetazoanALGs,weusedthereciprocalbest blasthit(RBH)orthologycriterioninanalignmentofthe Ixodesscapularis predictedproteins[36]totheconsensus sequencesforthemarkerscaffolds. L.polyphemus scaffolds withRBHofe-valuewereassignedtothesameancestral bilateriangeneorthologygroupastheir I.scapularis ortholog,andtherebywithhumangenes.Regionsofthemap weretestedforenrichmentingenesfromparticularancestrallinkagegroupswithFisher ’ sExactTest,andbreakpointsinancestrallinkagegroupcompositionwere identifiedusingahiddenMarkovmodel,aspreviously described[6,7].HomeoboxgenemodelingWeidentified155markerscaffo ldswithatblastxalignment ofe-valuetoasetofcheliceratehomeoboxgenesequences downloadedfromGenbankusingtheNCBIonlinequery interface(genbankaccessionsAF071402.1,AF071403.1, AF071405.1,AF071406.1,AF071407.1,AF085352.1,AF151 986.1,AF151987.1,AF151988.1,AF151989.1,AF151990.1, AF151991.1,AF151992.1,AF151993.1,AF151994.1,AF151 995.1,AF151996.1,AF151997.1,AF151998.1,AF151999.1, AF152000.1,AF237818.1,AJ005643.1,AJ007431.1,AJ00743 2.1,AJ007433.1,AJ007434.1, AJ007435.1,AJ007436.1,AJ007 437.1,AM419029.1,AM419030.1,AM419031.1,AM41903 2.1,DQ315728.1,DQ315729.1,DQ315730.1,DQ315731.1, DQ315732.1,DQ315733.1,DQ315734.1,DQ315735.1,DQ3 15736.1,DQ315737.1,DQ315738.1,DQ315739.1,DQ31574 0.1,DQ315741.1,DQ315742.1,DQ315743.1,DQ315744.1, (Seefigureonpreviouspage.) Figure9 Phylogenetictreeofhomeoboxsequences,part2. TherootedsubtreeprunedfromthetreeinFigure8 Nodesarelabeledwith Bayesianposteriorprobabilities.Highlysupportedpartitionsusedtoclassify L.polyphemus sequencesaredrawninred,withtheabbreviationfor theclassshowninlargeletters. L.polyphemus homeoboxsequencesnotgroupedintooneofthesehighlysupportedpartitionsareassignedto class “ hox? ” Table3MixturemodelfitstoKsdistributionNkln(L)BICAICMixturecomponents 12 273.93559.74551.871.320.50 25 259.32 548.31 528.640.700.14;1.450.45 38 253.36554.19522.710.710.17;1.340.29;2.090.19 411 251.41568.11524.820.740.18;1.340.20;1.700.04;2.020.22Nisthenumberofmixturecomponents,kthenumbermodelparameters,ln(L)theloglikelihoodofthedataunderthebestfitmodel. BIC,Bayesianinformationcriterion;AIC,Akaikeinformationcriterion. TheBICscoreofourselectedmodelisshowninbold.Nossa etal.GigaScience 2014, 3 :9 Page17of21 http://www.gigasciencejournal.com/content/3/1/9

PAGE 18

EU870887.1,EU870888.1,EU870889.1,HE608680.1,HE608 681.1,HE608682.1,HE805493.1,HE805494.1,HE805495.1, HE805496.1,HE805497.1,HE805498.1,HE805499.1,HE80 5500.1,HE805501.1,HE805502.1,S70005.1,S70006.1,S700 08.1,andS70010.1).Thereadsofeachmarker(those withbeststampy[61]alignmenttothescaffold)were reassembledwithPHRAP[65],withdefaultparameters.The resultingcontigswerealignedtoacollectionofhomeoboxcontainingproteinsequences(genbankaccessionsNP 001034497.1,NP001034510.1,AAL71874.1,NP001034505.1, NP001036813.1,CAA66399.1,NP001107762.1,NP0011 07807.1,EEZ99256.1,NP001034519.1,NP476954.1,NP032 840.1,NP031699.2,AAI37770.1,EEN68949.1,NP523700.2, NP001034511.2,AAK16421.1,andAAK16422.1)with exonerate[66]inprotein-to-genomemode.Foreach contig,theaminoacidsequencepredictedbythe highest-scoringexoneratealignmentwasusedinsubsequentphylogeneticanalysis,resultingin104putative homeobox-containingmarkersranginginlengthfrom 18to147aminoacids.PhylogeneticanalysisofhomeoboxgenesAmultiplesequencealignmentofthepredictedhomeoboxsequencescombinedwithacollectionofrepresentativesequencesfromvariousclassesofhomeoboxgenes wasconstructedwithmusclev3.8.31[67]usingdefault settings.Theresultingalignmentwastrimmedtoa63 aminoacidsegmentspanningtheconservedhomeodomain,andsequenceswithmorethan50%gapswere removed,leaving93predicted L.polyphemus homeobox genesintheanalysis.Bayesianphylogeneticanalysiswas carriedoutontheresulting178taxon,63aminoacid charactermatrix(SeeAdditionalfile1)usingMrBayes v3.2.1[68]usingamixedmodelofaminoacidsubstituions,gamma-distributedratevariationamongsiteswith fixedshapeparameter =1.0,alignmentgapstreatedas missingdata,2,000,000MonteCarlosteps,twoindependentrunswithfourMonteCarlochains,andtheinitial 25%ofsampledtreeswerediscardedas “ burn-in ” .Monte Carloappearedtoreachconvergence,withanaverage standarddeviationofthesplitfrequenciesof0.022.The majority-ruleconsensusofthesampledtreesisshownin Figures8and9,andwell-supportedgeneclades(posterior probabilitygreaterthan0.95)wereusedtogroupthe predicted L.polyphemus genesintoclasses.Thetablein Additionalfile1liststhereassembledmarkercontigs, theirinferredhoxgeneclass,andmaximumlikelihood mappositions.Predictedgeneswereanchoredtothemap asdescribedabove.GenomicdistributionofparalogsWeidentified2,716pairsof Limulus markersthatcanboth beplacedonthemapandhavetheirbesttranslatedalignmenttothesame Ixodesscapularis gene.( I.scapularis geneswithmorethanfivebest-hitmarkerswereexcluded fromseedingsuchpairs.)Toestimatethesynonymoussequencedivergencebetweenpairsofcandidate L.polyphemus paralogousgenepairsand L.polyphemus genesand their T.tridentatus orthologs,weconstructedcodonalignmentsofpredictedcodingsequenceforestimationofsynonymoussequencedivergence.Conservedclustersof paralogswereidentifiedusingavariantofthe “ max-gap ” Figure10 TwocomponentmixturemodelfittotheKspeakontherange0<=Ks<=2.5. Thebest-fittingmodelwasselectedbytheBayesian InformationCriterion(Table3).Thecomponentmeansare0.7and1.45substitutionspersite.ThepositionofthepeakatlowestKswasnotsensitiveto theadditionofmoremixturecomponents. Nossa etal.GigaScience 2014, 3 :9 Page18of21 http://www.gigasciencejournal.com/content/3/1/9

PAGE 19

criterion[3]inwhichtwogenesareplacedinthesame clusteriftheyandtheirparalogsliewithinthreshold distance.KaandKsestimationforparalogsandT.tridentatus orthologsFigure6showsthedistributionacrossthemapofpairsof candidateparalogs.Toestimatethesynonymoussequence divergencebetweenpairsofcandidate L.polyphemus paralogousgenepairs,andbetween L.polyphemus genesand their T.tridentatus orthologs,wefollowedthefollowing protocol. 1.ReassemblereadsfromeachmarkerwithPHRAP [ 65 ],andcreateapredictedcodingsequenceusing exonerate,asdescribedfortheannotationof homeoboxgenemodels(seeabove). 2.Combinetheexoneratealignmentsofcodonsto aminoacidstocreateanalignmentofcodonsfor eitherapairof L.polyphemus sequences,orfora L.polyphemus T.tridentatus sequencepair. 3.UsethemethodofYangandNielsen[ 69 ]to estimatethesynonymousandnon-synonymous substitutionratesKaandKs,asimplementedinthe KaKsCalculatorpackage[ 70 ]. 4.Discardestimatesbasedonfewerthan30sites (30synonymoussitesforestimatesofKs, non-synonymoussitesforKa). GenBankaccessionsfor Tachypleustridentatus mRNA clones:JQ966943,AB353281,AB353280,HM156111,HQ22 1882,HQ221883,HQ221881,HQ386702,HM852953,TAT TPP,TATPROCLOT,FN582225,FN582226,AF467804,AF 227150,GQ260127,AF264067,AF264068,AB353279,AB0 05542,TATLICI,TATTGL,TATCFGB,TATLFC1,TATLF C2,AB201713,TATCFGA,TATLICI2,CS423581,CS423579, AB028144,AB201778,AB201776,AB201774,AB201772,AB 201770,AB201768,AB201766,AB201779,AB201777,AB20 1775,AB201773,AB201771,AB201769,AB201767,AB201 765,AB105059,AB002814,AX763473,TATCFBP,AB076 186,AB076185,X04192,TATHCLL,AB037394,AB019116, AB019114,AB019112,AB019110,AB019108,AB019106,AB 019104,AB019102,AB019100,AB019098,AB019096,AB 019117,AB019115,AB019113,AB019111,AB019109,AB 019107,AB019105,AB019103,AB019101,AB019099,AB 019097,AB023783,AB024738,AB024739,AB024737,AB0 17484,D87214,D85756,D85341. Figure7showsthedistributionofKaandKsforparalogs and T.tridentatus orthologs.Toestimatethenumberand ageofpeaksintheun-saturatedrange[37]oftheKsdistribution(andofputativeWGDevents),wefitaseriesof univariatenormalmixturemodels,with1,2,3,and4 componentstotheparalogKsdistributionintherange 0<=Ks<=2.5andselectedthebestmodelonthebasis ofBayesianInformationCriterion(BIC)(Table3).The bestmodelhadtwocomponents,withmeansat0.7and 1.45substitutionspersite.Thepositionofthepeakat lowestKswasnotsensitivetotheadditionofmore mixturecomponents.Figure10showscomparisonof thedistributionandthecomponentsofthebest-fitting model.GaussianmixturemodelswereestimatedinR withmixtools[71].AvailabilityofsupportingdataTherawsequencingreadsarecurrentlybeingsubmittedthroughtheNCBISRAandareaccessibleviaNCBI BioProjectaccessionPRJNA187356. Thedatasetssupportingtheresultsofthisarticleare availableinthe GigaScience GigaDBrepository[72].AdditionalfileAdditionalfile1: Theaminoacidcharactermatrixusedforthe phylogeneticanalysisofhomeoboxgenes. Abbreviations ALG: Ancestrallinkagegroups;AIC:Akaikeinformationcriterion;BIC:Bayesian informationcriterion;bp:Basepair;EM:Expectationmaximization; GBS:Genotyping-by-sequencing;JAM:Jointassemblyandmapping; LG:Linkagegroup;Max-gap:Maximumgap;Mb:Megabase;MYA:Million yearsago;PE:Paired-end;RAD-Seq:RestrictionsiteassociatedDNA sequencing;RBH:Reciprocalbestblasthit;SNP:Singlenucleotide polymorphism;WGD:Whole-genomeduplication. Competinginterests Theauthorsdeclarethattheyhavenocompetinginterests. Authors ’ contributions NHPconceivedandledtheproject.Allauthorswrotethepaper.PH,NHP andJ-XYwrotesoftware.NHP,PH,J-XYandCWNcarriedoutsequence analysis.JBcollectedandraisedsamples.CWNextractedgenomicDNAand createdthelibrariesforsequencing.Allauthorsreadandapprovedthefinal manuscript. Acknowledgements ThisresearchwassupportedbytheNationalScienceFoundation (EF-0850294andIOB06 – 41750),theBeckmanYoungInvestigatorProgram, theUniversityofFloridaDivisionofSponsoredResearch,theDepartmentof Biology,andtheUFMarineLaboratoryatSeahorseKey.Wethankthethree reviewers,Dr.HuguesRoestCrollius,Dr.StephenRichardsandDr.BrianEads, fortheirvaluablecommentswhichgreatlyhelpedtoimprovethequalityof thispaper. Authordetails1DepartmentofEcologyandEvolutionaryBiology,RiceUniversity,P.O.Box 1892,Houston,TX77251-1892,USA.2DepartmentofBiochemistryandCell Biology,RiceUniversity,P.O.Box1892,Houston,TX77251-1892,USA.3DepartmentofBiology,UniversityofFlorida,P.O.Box11-8525Gainesville,FL 32611-8525,USA.4Currentaddress:GenebyGene,Ltd,Houston,TX77008, USA. Received:23October2013Accepted:23April2014 Published:14May2014 References1.OhnoS: EvolutionbyGeneDuplication. Springer-Verlag;1970. 2.WolfeKH,ShieldsDC: Molecularevidenceforanancientduplicationof theentireyeastgenome. Nature 1997, 387: 708 – 713.Nossa etal.GigaScience 2014, 3 :9 Page19of21 http://www.gigasciencejournal.com/content/3/1/9

PAGE 20

3.McLysaghtA,HokampK,WolfeKH: Extensivegenomicduplicationduring earlychordateevolution. NatGenet 2002, 31: 200 – 204. 4.SimillionC,VandepoeleK,MontaguMCEV,ZabeauM,dePeerYV: Thehidden duplicationpastofArabidopsisthaliana. ProcNatlAcadSci 2002, 99: 13627 – 13632. 5.AuryJ-M,JaillonO,DuretL,NoelB,JubinC,PorcelBM,SgurensB,Daubin V,AnthouardV,AiachN,ArnaizO,BillautA,BeissonJ,BlancI,BouhoucheK, CmaraF,DuharcourtS,GuigoR,GogendeauD,KatinkaM,KellerA-M, KissmehlR,KlotzC,KollF,MoulAL,LepreG,MalinskyS,NowackiM, NowakJK,PlattnerH, etal : Globaltrendsofwhole-genomeduplications revealedbytheciliateParameciumtetraurelia. Nature 2006, 444: 171 – 178. 6.PutnamNH,SrivastavaM,HellstenU,DirksB,ChapmanJ,SalamovA,Terry A,ShapiroH,LindquistE,KapitonovVV,JurkaJ,GenikhovichG,GrigorievIV, LucasSM,SteeleRE,FinnertyJR,TechnauU,MartindaleMQ,RokhsarDS: Seaanemonegenomerevealsancestraleumetazoangenerepertoire andgenomicorganization. Science 2007, 317: 86 – 94. 7.PutnamNH,ButtsT,FerrierDEK,FurlongRF,HellstenU,KawashimaT, Robinson-RechaviM,ShoguchiE,TerryA,YuJ-K,Benito-GutierrezE, DubchakI,Garcia-FernandezJ,Gibson-BrownJJ,GrigorievIV,HortonAC, deJongPJ,JurkaJ,KapitonovVV,KoharaY,KurokiY,LindquistE,LucasS, OsoegawaK,PennacchioLA,SalamovAA,SatouY,Sauka-SpenglerT, SchmutzJ,Shin-IT, etal : Theamphioxusgenomeandtheevolutionof thechordatekaryotype. Nature 2008, 453: 1064 – 1071. 8.SrivastavaM,BegovicE,ChapmanJ,PutnamNH,HellstenU,KawashimaT, KuoA,MitrosT,SalamovA,CarpenterML,SignorovitchAY,MorenoMA, KammK,GrimwoodJ,SchmutzJ,ShapiroH,GrigorievIV,BussLW, SchierwaterB,DellaportaSL,RokhsarDS: TheTrichoplaxgenomeandthe natureofplacozoans. Nature 2008, 454: 955 – 960. 9.SrivastavaM,SimakovO,ChapmanJ,FaheyB,GauthierMEA,MitrosT, RichardsGS,ConacoC,DacreM,HellstenU,LarrouxC,PutnamNH,Stanke M,AdamskaM,DarlingA,DegnanSM,OakleyTH,PlachetzkiDC,ZhaiY, AdamskiM,CalcinoA,CumminsSF,GoodsteinDM,HarrisC,JacksonDJ, LeysSP,ShuS,WoodcroftBJ,VervoortM,KosikKS, etal : TheAmphimedon queenslandicagenomeandtheevolutionofanimalcomplexity. Nature 2010, 466: 720 – 726. 10.SimakovO,MarletazF,ChoS-J,Edsinger-GonzalesE,HavlakP,HellstenU, KuoD-H,LarssonT,LvJ,ArendtD,SavageR,OsoegawaK,deJongP, GrimwoodJ,ChapmanJA,ShapiroH,AertsA,OtillarRP,TerryAY,BooreJL, GrigorievIV,LindbergDR,SeaverEC,WeisblatDA,PutnamNH,RokhsarDS: Insightsintobilaterianevolutionfromthreespiraliangenomes. Nature 2013, 493: 526 – 531. 11.LvJ,HavlakP,PutnamN: Constraintsongenesshapelong-termconservation ofmacro-syntenyinmetazoangenomes. BMCBioinformatics 2011, 12 (Suppl9):S11. 12.EarlD,BradnamK,JohnJS,DarlingA,LinD,FassJ,YuHOK,BuffaloV, ZerbinoDR,DiekhansM,NguyenN,AriyaratnePN,SungW-K,NingZ, HaimelM,SimpsonJT,FonsecaNA,Birol ,DockingTR,HoIY,RokhsarDS,ChikhiR,LavenierD,ChapuisG,NaquinD,MailletN,SchatzMC,KelleyDR, PhillippyAM,KorenS, etal : Assemblathon1:Acompetitiveassessmentof denovoshortreadassemblymethods. GenomeRes 2011, 21: 2224 – 2241. 13.DaveyJW,HohenlohePA,EtterPD,BooneJQ,CatchenJM,BlaxterML: Genome-widegeneticmarkerdiscoveryandgenotypingusing next-generationsequencing. NatRevGenet 2011, 12: 499 – 510. 14.AltshulerD,PollaraVJ,CowlesCR,VanEttenWJ,BaldwinJ,LintonL,Lander ES: AnSNPmapofthehumangenomegeneratedbyreduced representationshotgunsequencing. Nature 2000, 407: 513 – 516. 15.BairdNA,EtterPD,AtwoodTS,CurreyMC,ShiverAL,LewisZA,SelkerEU, CreskoWA,JohnsonEA: RapidSNPdiscoveryandgeneticmappingusing sequencedRADmarkers. PLoSONE 2008, 3: e3376. 16.HuangX,FengQ,QianQ,ZhaoQ,WangL,WangA,GuanJ,FanD,Weng Q,HuangT,DongG,SangT,HanB: High-throughputgenotypingby whole-genomeresequencing. GenomeRes 2009, 19: 1068 – 1076. 17.AndolfattoP,DavisonD,ErezyilmazD,HuTT,MastJ,Sunayama-MoritaT, SternDL: Multiplexedshotgungenotypingforrapidandefficientgenetic mapping. GenomeRes 2011, 21: 610 – 617. 18.ChapmanJA,HoI,SunkaraS,LuoS,SchrothGP,RokhsarDS: Meraculous:De novogenomeassemblywithshortpaired-endreads. PLoSONE 2011, 6: e23501. 19.ButlerJ,MacCallumI,KleberM,ShlyakhterIA,BelmonteMK,LanderES, NusbaumC,JaffeDB: ALLPATHS:Denovoassemblyofwhole-genome shotgunmicroreads. GenomeRes 2008, 18: 810 – 820. 20.GnerreS,MacCallumI,PrzybylskiD,RibeiroFJ,BurtonJN,WalkerBJ,Sharpe T,HallG,SheaTP,SykesS,BerlinAM,AirdD,CostelloM,DazaR,WilliamsL, NicolR,GnirkeA,NusbaumC,LanderES,JaffeDB: High-qualitydraft assembliesofmammaliangenomesfrommassivelyparallelsequence data. ProcNatlAcadSci 2011, 108: 1513 – 1518. 21.RudkinDM,YoungGA: Horseshoecrabs – anancientancestryrevealed. In BiolConservHorseshoeCrabs. EditedbyTanacrediJT,BottonML,SmithD. NewYork:Springer;2009:25 – 44. 22.FisherDC: TheXiphosurida:archetypesofBradytely? In LivingFoss.Edited byEldregeN,StanleySM.NewYork:Springer;1984:196 – 213. 23.BerksonJ,ShusterCNJr: Thehorseshoecrab:thebattleforatrue multiple-useresource. Fisheries 1999, 24: 6 – 10. 24.ShusterCNJr,BarlowRB,BrockmannHJ(Eds): TheAmericanHorseshoeCrab. Cambridge,MA:HarvardUniversityPress;2003. 25.GregoryTR: AnimalGenomeSizeDatabase. http://www.genomesize.com. 26.PopM,KosackDS,SalzbergSL: Hierarchicalscaffoldingwithbambus. GenomeRes 2004, 14: 149 – 159. 27.VanOsH,AndrzejewskiS,BakkerE,BarrenaI,BryanGJ,CaromelB,Ghareeb B,IsidoreE,deJongW,vanKoertP,LefebvreV,MilbourneD,RitterE,van derVoortJNAMR,Rousselle-BourgeoisF,vanVlietJ,WaughR,VisserRGF, BakkerJ,vanEckHJ: Constructionofa10,000-markerultradensegenetic recombinationmapofpotato:providingaframeworkforaccelerated geneisolationandagenomewidephysicalmap. Genetics 2006, 173: 1075 – 1087. 28.SekiguchiK: BiologyofHorseshoeCrabs. Tokoyo:ScienceHouseCo.,Ltd.; 1988:50 – 68. 29.CartwrightRA,HussinJ,KeeblerJEM,StoneEA,AwadallaP: Afamily-based probabilisticmethodforcapturingdenovomutationsfromhighthroughputshort-readsequencingdata. StatApplGenetMolBiol 2012, 11 (2). 30.BirdAP: DNAmethylationandthefrequencyofCpGinanimalDNA. NucleicAcidsRes 1980, 8: 1499 – 1504. 31.LynchM: Theoriginsofeukaryoticgenestructure. MolBiolEvol 2006, 23: 450 – 468. 32.BromanKW,MurrayJC,SheffieldVC,WhiteRL,WeberJL: Comprehensive humangeneticmaps:individualandsexspecificvariationinrecombination. AmJHumGenet 1998, 63: 861 – 869. 33.KongA,GudbjartssonDF,SainzJ,JonsdottirGM,GudjonssonSA,Richardsson B,SigurdardottirS,BarnardJ,HallbeckB,MassonG,ShlienA,PalssonST,Frigge ML,ThorgeirssonTE,GulcherJR,StefanssonK: Ahigh-resolution recombinationmapofthehumangenome. NatGenet 2002, 31: 241 – 247. 34.CoopG,PrzeworskiM: Anevolutionaryviewofhumanrecombination. NatRevGenet 2007, 8: 23 – 34. 35.HellmannI,EbersbergerI,PtakSE,PboS,PrzeworskiM: Aneutral explanationforthecorrelationofdiversitywithrecombinationratesin humans. AmJHumGenet 2003,72: 1527 – 1535. 36.VectorBase: Ixodesscapularisannotation,IscaW1. https://www.vectorbase. org/organisms/ixodes-scapularis. 37.VannesteK,VandePeerY,MaereS: Inferenceofgenomeduplications fromagedistributionsrevisited. MolBiolEvol 2013, 30: 177 – 190. 38.ObstM,FaurbyS,BussarawitS,FunchP: Molecularphylogenyofextant horseshoecrabs(Xiphosura,Limulidae)indicatesPaleogene diversificationofAsianspecies. MolPhylogenetEvol 2012, 62: 21 – 26. 39.BegunDJ,AquadroCF: Levelsofnaturallyoccurring DNApolymorphismcorrelate withrecombinationratesinD.melanogaster. Nature 1992, 356: 519 – 520. 40.CutterAD,PayseurBA: Selectionatlinkedsitesinthepartialselfer caenorhabditiselegans. MolBiolEvol 2003, 20: 665 – 673. 41.NachmanMW: Singlenucleotidepolymorphismsandrecombinationrate inhumans. TrendsGenet 2001, 17: 481 – 485. 42.StephanW,LangleyCH: DNApolymorphisminlycopersiconand crossing-overperphysicallength. Genetics 1998, 150: 1585 – 1593. 43.RoseliusK,StephanW,StdlerT: Therelationshipofnucleotide polymorphism,recombinationrateandselectioninwildtomatospecies. Genetics 2005, 171: 753 – 763. 44.AndolfattoP,PrzeworskiM: Regionsoflowercrossingoverharbormore rarevariantsinAfricanpopulationsofdrosophilamelanogaster. Genetics 2001, 158: 657 – 665. 45.HavlakP,ChenR,DurbinKJ,EganA,RenY,SongX-Z,WeinstockGM,Gibbs RA: Theatlasgenomeassemblysystem. GenomeRes 2004, 14: 721 – 732. 46.TreangenTJ,SommerDD,AnglyFE,KorenS,PopM: Nextgenerationsequence assemblywithAMOS. CurrentProtocolsinBioinformatics 2011, 33: 11.8.1 – 11.8.18. 47.TheJAM-pipeline. https://github.com/putnamlab/jam-pipeline. 48.JohnsonSL,BrockmannHJ: Costsofmultiplemates:anexperimental studyinhorseshoecrabs. AnimBehav 2010, 80: 773 – 782.Nossa etal.GigaScience 2014, 3 :9 Page20of21 http://www.gigasciencejournal.com/content/3/1/9

PAGE 21

49.MullikinJC,NingZ: Thephusionassembler. GenomeRes 2003, 13: 81 – 90. 50.CockPJA,FieldsCJ,GotoN,HeuerML,RicePM: TheSangerFASTQfile formatforsequenceswithqualityscores,andtheSolexa/IlluminaFASTQ variants. NucleicAcidsRes 2010, 38: 1767 – 1771. 51.EwingB,GreenP: Base-callingofautomatedsequencertraces usingphred.ii.errorprobabilities. GenomeRes 1998, 8: 186 – 194. 52.KnuthDE: SearchingandSorting ,Volume3.Reading,MA:Addison-Wesley; 1973[ TheArtofComputerProgramming ]. 53.RobertsM,HayesW,HuntBR,MountSM,YorkeJA: Reducingstoragerequirements forbiologicalsequencecomparison. Bioinformatics 2004, 20: 3363 – 3369. 54.YeC,MaZS,CannonCH,PopM,YuDW: Exploitingsparsenessindenovo genomeassembly. BMCBioinformatics 2012, 13 (6) : S1. 55. Ensembl. http://www.ensembl.org/index.html. 56.ChenGK,MarjoramP,WallJD: FastandflexiblesimulationofDNA sequencedata. GenomeRes 2009, 19: 136 – 142. 57.DehalP,SatouY,CampbellRK,ChapmanJ,DegnanB,TomasoAD, DavidsonB,GregorioAD,GelpkeM,GoodsteinDM,HarafujiN,Hastings KEM,HoI,HottaK,HuangW,KawashimaT,LemaireP,MartinezD, MeinertzhagenIA,NeculaS,NonakaM,PutnamN,RashS,SaigaH,Satake M,TerryA,YamadaL,WangH-G,AwazuS,AzumiK, etal : Thedraft genomeofcionaintestinalis:insightsintochordateandvertebrate origins. Science 2002, 298: 2157 – 2167. 58.HauboldB,PfaffelhuberP,LynchM: mlRho – aprogramforestimatingthe populationmutationandrecombinationratesfromshotgun-sequenced diploidgenomes. MolEcol 2010, 19: 277 – 284. 59.SmallKS,BrudnoM,HillMM,SidowA: Extremegenomicvariationina naturalpopulation. ProcNatlAcadSci 2007, 104: 5698 – 5703. 60. dwgsim. https://github.com/nh13/DWGSIM/releases. 61.LunterG,GoodsonM:Stampy:astatisticalalgorithmforsensitiveand fastmappingofIlluminasequencereads. GenomeRes 2011, 21: 936 – 939. 62.LiH,HandsakerB,WysokerA,FennellT,RuanJ,HomerN,MarthG,Abecasis G,DurbinR: Thesequencealignment/MapformatandSAMtools. Bioinformatics 2009, 25: 2078 – 2079. 63.YangZ: StatisticalpropertiesofaDNAsampleunderthefinite-sites model. Genetics 1996, 144: 1941 – 1950. 64.HaldaneJBS: Thecombinationoflinkagevaluesandthecalculationof distancesbetweenthelocioflinkedfactors. JGenet 1919, 8: 299 – 309. 65.GordonD,AbajianC,GreenP: Consed:agraphicaltoolforsequence finishing. GenomeRes 1998, 8: 195 – 202. 66.SlaterGS,BirneyE: Automatedgenerationofheuristicsforbiological sequencecomparison. BMCBioinformatics 2005, 6: 31. 67.EdgarRC: MUSCLE:multiplesequencealignmentwithhighaccuracyand highthroughput. NucleicAcidsRes 2004, 32: 1792 – 1797. 68.RonquistF,HuelsenbeckJP: MrBayes3:Bayesianphylogeneticinference undermixedmodels. Bioinformatics 2003, 19: 1572 – 1574. 69.YangZ,NielsenR: Estimatingsynonymousandn onsynonymoussubstitution ratesunderrealisticevolutionarymodels. MolBiolEvol 2000, 17: 32 – 43. 70.ZhangZ,LiJ,ZhaoX-Q,WangJ,WongGK-S,YuJ: KaKscalculator:calculating KaandKsthroughmodelselectionandmodelaveraging. Genomics ProteomicsBioinformatics 2006, 4: 259 – 263. 71.BenagliaT,ChauveauD,Hunter,DavidR,Young,DerekS: mixtools:AnR packageforanalyzingfinitemixturemodels. JStatSoftw 2009, 32: 1 – 29. 72.NossaCW,HavlakP,YueJX,LvJ,VincentKY,BrockmannHJ,PutnamNH: Supportingmaterialsfrom “Jointassemblyandgeneticmappingofthe Atlantichorseshoecrabgenomerevealsancientwholegenome duplication ” 2014.GigaScienceDatabase.http://dx.doi.org/10.5524/100091.doi:10.1186/2047-217X-3-9 Citethisarticleas: Nossa etal. : Jointassemblyandgeneticmappingof theAtlantichorseshoecrabgenomerevealsancientwholegenome duplication. GigaScience 2014 3 :9. Submit your next manuscript to BioMed Central and take full advantage of: € Convenient online submission € Thorough peer review € No space constraints or color “gure charges € Immediate publication on acceptance € Inclusion in PubMed, CAS, Scopus and Google Scholar € Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit Nossa etal.GigaScience 2014, 3 :9 Page21of21 http://www.gigasciencejournal.com/content/3/1/9



PAGE 1

name sequence accession position class '.' for match LG:coord. EVX-1 QMRRYRTAFTREQIARLEKEFYRENYVSRPRRCELAAALNLPETTIKVWFQNRRMKDKRQRLA P23683.1 g_ayrxn_scaff_4445_1.Contig24 --...........L.......C..S..................S....R.------------1 : 7968 evx g_algxc_scaff_4707_1.Contig9 KE..S.....E..LEY..RT.E.TQ.PDVYT.E...QRTK.T.ARVQ.--------------1 : 2335 ? g_ajgny_scaff_5092_1.Contig4 KAK.I..I..A..LN...A..LQQQ.MVGRE.SH..SE...K.AQV........I.WRK---4 : 26846 ? g_ahyur_scaff_4335_1.Contig2 -AK.V..I..A..LD...T..E.QQ.MVG.E.LQ..S....T.AQV........I.WRK.HFD 1 : 3106 ? engrailed__Tribolium_castaneum_ EEK.P....SGA.L...KH..AENR.LTER..QQ.S.E.G.N.AQ..I....K.A.I.KASGQ NP_001034511.2 engrailed__isoform_A__Drosophila_melanogaster_ DEK.P....SS..L...KR..NENR.LTER..QQ.SSE.G.N.AQ..I....K.A.I.KSTGS NP_523700.2 engrailed__Branchiostoma_oridae_ EEK.P.....S..LQ..K...QENR.LTEQ..QD..RE.K.N.SQ..I....K.A.I.KAAGV EEN68949.1 g_addne_scaff_4497_1.Contig54 EEK.P.....TD.L...K...HENR.LTEK..Q...RE.Q.N.SQ..I....K.A.I.KSTGE 1 : 5794 en g_ausca_scaff_4858_1.Contig1 DEK.P.....AD.L...KQ..QENR.LTEK..QD..LD.Q.N.SQ..I....K.A.I.KASGQ 20 : 83001 en g_adwrx_scaff_4421_1.Contig72 DEK.P.....AD.L...KA..QENR.LTEK..QD..RG.Q.N.SQ..I....K.A.I.KATGH 15 : 71297 en g_avzuc_scaff_4437_1.Contig3 DEK.P.....TD.L...KA..QENK-LTEK..QD..RE------------------------21 : 85367 en g_adwrx_scaff_4421_1.Contig61 DEK.P.....TD.L...KA..QENR.LTEK..QD..RE.Q.N.SQ..I.-------------15 : 71297 en g_avzuc_scaff_4437_1.Contig4 -------------------..QENR.LTEK..QD..RE.Q.N.SQ..I....K.A.I.KATGH 21 : 85367 en g_adwrx_scaff_4421_1.Contig63 -EK.P.....AD.L...KA..QENR.LTEK..QD..RE.Q.N.SQ..I....K.A.I.KAVGH 15 : 71297 en g_ayadw_scaff_4575_1.Contig4 DEK.P.....AD.L...KA..QENR.LTEK..QD..RE.Q.N.SQ..I....K.A.I.KAVGH 3 : 22654 en Nkx-2.2a KK.KR.VL.SKA.TYE..RR.RQQR.L.A.E.EH..SLIR.TP.QV.I....H.Y.M..A.AE NP_571497.1 Nkx-2.2 KK.KR.VL.SKA.TYE..RR.RQQR.L.A.E.EH..SLIR.TP.QV.I....H.Y.M..A.AE NP_002500.1 Nkx-2.2__Gallus_gallus_ KK.KR.VL.SKA.TYE..RR.RQQR.L.A.E.EH..SLIR.TP.QV.I....H.Y.M..A.AE XP_003643855.1 g_ajlua_scaff_4446_1.Contig10 KK.KR.VL.SKA.TYE..RR.RQQR.L.A.E.EH..NFIR.TP.QV.I....H.Y.T..A.QE 1 : 1567 nkx2 g_acfsj_scaff_4532_1.Contig25 KK.KR.VL.SKA.TYE..RR.RQQR.L.A.E.EH..SIIR.TP.QV.I....H.Y.T..A.QE 3 : 20058 nkx2 ladybird-like__3 KR.KS....SNQ.LFE..RR.V.QK.L.PAD.DQ..QR.A.SSAQVIT......A.L..DLDE NP_001038133.1 lady_bird-like__1_homolog KR.KS.....NH..YE...R.LYQK.L.PAD.DQI.QQ.G.TNAQVIT......A.L..DLEE NP_001007135.1 g_acdli_scaff_4927_1.Contig5 KK.KS.....NH..FE...R.LYQK.L.PAD.D.I.QS.G.TNAQVIT......A.L..---26 : 93066 lbx g_agfxy_scaff_4755_1.Contig8 KK.KS.....NQ..FE...R.LYQK.L.PAD.D.I.QS.R.TNAQVIT......A.L..---13 : 67147 lbx g_aplvp_scaff_4845_1.Contig14 KK.KS.....NH..FE...R.LYQK.L.PAD.D.I.QS.G.TNAQVIT......A.L..---1 : 3106 lbx AGAP003673PA__Anopheles_gambiae_str._PEST_ KK.KS.....NH..FE...R.LYQK.L.PSD.D.I....G.SNAQVIT......A.L..DMEE EAA04094 g_amhlt_scaff_4436_1.Contig3 KQ.KA.....DH.LQT...C.E.QK.L.VQE.M...SK...TD.QV.T.Y....--------8 : 46729 ? g_amhlt_scaff_4436_1.Contig8 KE.KA.....DY..QT...R.QTQK.L.VQD.M.I.SK...TD.HV.T.Y....--------8 : 46729 ? g_aeqlo_scaff_4727_1.Contig6 RKKKT..V...S.VFQ..ST.DMKR.L.SSE.AG...C.H.T..QV.I......N.W...--22 : 86973 ? g_agfda_scaff_4460_1.Contig8 KRKKS..T..GR..FE...Q.EIKK.L.SSE.A.M.KL..VT.QQVEI--------------26 : 93149 ? g_artzi_scaff_4429_1.Contig73 SN.KL..P..TQ.LLA..RK.LAKQ.L.IAE.AKFSIS.T.T.AQV.I......A.E..---5 : 32215 msx Msx__Saccoglossus_kowalevskii_ TN.KP..P..TS.LLA..RK.RQKQ.L.IAE.A.FS.S...T..QV.I......A.A..LQE. ABD97280 Msx___Nematostella_vectensis_ AN.KP..P..TS.LLA..RK.RQKQ.L.IAE.A.FS.S...T..QV.I......A.A..LHE. BAG11598 MSX-2__Mus_musculus_ TN.KP..P..TS.LLA..RK.RQKQ.L.IAE.A.FSSS...T..QV.I......A.A..LQE. NP_038629 MSX-2__Homo_sapiens_ TN.KP..P..TS.LLA..RK.RQKQ.L.IAE.A.FSSS...T..QV.I......A.A..LQE. BAA06549 MSX1___Homo_sapiens_ TN.KP..P..TA.LLA..RK.RQKQ.L.IAE.A.FSSS.S.T..QV.I......A.A..LQE. NP_002439 MSX1_variant__Mus_musculus_ TN.KP..P..TA.LLA..RK.RQKQ.L.IAE.A.FSSS.S.T..QV.I......A.A..LQE. AAG32466 g_aebwc_scaff_4685_2.Contig8 -----------------------KQ.L.IAE.A.FSSS.S.T..QV.I......A.E..LKE. 3 : 19118 msx g_aebwc_scaff_4685_2.Contig10 SN.KP..P..TQ.LLA..RK.RTKQ.L.IAE.A.FSSS...T..QV.I......A.E..---3 : 19118 msx g_aebwc_scaff_4685_2.Contig13 SN.KP..P..TQ.LLA..RK.RSKQ.L.IAE.A.FSSS...T..QV.I......A.E..---3 : 19118 msx g_aoopl_scaff_4606_1.Contig8 -----------------------KQ.L.IAE.A.FSSS...T..QV.I......A.EN.LKE26 : 93149 msx g_aoopl_scaff_4606_1.Contig18 SS.KP..P..TK.LLA...M.QTKQ.L.IAE.A.FSSSM..T..QV.I......A.E..---26 : 93149 msx g_agizq_scaff_4515_1.Contig28 --..R...Y.S..LLQ.....HAKK.L.LTE.SQISTT.Q.S.VQV.I......A.W..---3 : 23600 gbx unplugged__Drosophila_melanogaster_ KS..R.....S..LLE..R..HAKK.L.LTE.SQI.TS.K.S.VQV.I......A.W..VKAG NP_477146 AGAP006923PA__Anopheles_gambiae_str._PEST_ KS..R.....S..LLE..R..HAKK.L.LTE.SQI.TS.K.S.VQV.I......A.W..VKAG EAA04094 g_apstr_scaff_4456_1.Contig3 --..R.....N..LLE..N..HCKK.L.LTE.SQI.HS.Q.S.QQV.I......A.W..---1 : 8283 gbx GBX-1__Canis_lupus_familiaris_ KS..R.....S..LLE.....HCKK.L.LTE.SQI.H..K.S.VQV.I......A.W..IKAG XP_853664 le:///Users/yjx/Dropbox/gigasciencereviews/gigascience_R2_s... 1 of 4 4/22/14, 12:20 PM

PAGE 2

name sequence accession position class '.' for match LG:coord. g_aglhj_scaff_4667_1.Contig13 --..R.....S..LLE.....HSKK.L.LTE.SQI.HS.Q.S.VQV.I......A.W..---11 : 60310 gbx g_aglhj_scaff_4783_1.Contig4 KS..R.....S..LLE.....HSKK.L.LTE.SHI.NS.Q.S.VQV.I......A.W..VKAG 18 : 80043 gbx g_acgnb_scaff_4629_1.Contig3 RFLLSLSQR.SRNGSQATDGLHFNRWLT.R..I.IVRT.C.SD-------------------20 : 83755 hox? g_awtvf_scaff_4744_1.Contig2 ETK.Q...Y..H..LE.....HFNR.LT.R..I.I.HSSC.ERIK.----------------21 : 85243 hox? g_awtvf_scaff_4744_1.Contig1 -----------------------SR.LT.R.QI.I.HS.C.S.RQ..N...I....W.KDNKL 21 : 85243 hox? Drosophila-Dfd__NM_057853_ EPK.Q...Y..H..LE.....HYNR.LT.R..I.I.HT.V.S.RQ..I........W.KD--NM_057853 Tribolium-Dfd__EEZ99253_ EPK.Q...Y..H..LE.....HYNR.LT.R..I.I.HT.V.S.RQ..I........W.KD--EEZ99253 g_ajivj_scaff_4414_1.Contig4 ETK.Q...Y..H..LE.....HFNR.LT.R..I.I.HL.C.S.RQ..I........W.KDNKL 15 : 71079 hox? g_ajivj_scaff_4414_1.Contig3 ETK.Q...Y..H..LE.....HFNR.LT.R..I.I.HS.C.S.RQ..I........W.KDNKL 15 : 71079 hox? Branchiostoma-Hox4__AB028208_ DTK.S...Y..Q.VLE.....HFNR.LT.R..I.I.HS.G.T.RQ..I........W.KD--AB028208 g_akvhc_scaff_4400_4.Contig7 EAK.Q...Y..H.VLE.....HFNR.LT.R..I.I.H..C.S.RQ..I........W.KDNKL 18 : 79524 hox? Capitella-Dfd__EU196540_ DSK.T...Y..H..LE.....HFNR.LT.R..I.I.HT.C.S.RQ..I........W.KE--EU196540 Lottia-Dfd__110972_ DSK.N...Y..H.VLE.....HFNR.LT.R..I.I.HT.C.S.RQ..I........W.KE--JGI Lotgi1 protein id: 110972 Branchiostoma-Hox12__AAF81903_ SS.KK.CPYSKV.LLE.....LYNM.IT.EQ.G.I.RKV..TDRQV.I........M..M--AAF81903 Branchiostoma-Hox10__Z35150_ VG.KK.CPY.KY..LE.....LFNM....E..Q.ISRHV..SDRQV.I........M..M--Z35150 Branchiostoma-Hox11__AAF81909_ ST.KK.CPY.KY.TLE.....LFNMF.T.E..Q.I.RQ...TDRQV.I........M..M--AAF81909 Branchiostoma-Hox9__ABX39493_ SS.KK.CPY..F.TLE.....LYNM.LT.E..Y.ISQHV..T.RQV.I........M.KM--ABX39493 Branchiostoma-Hox13__AAF81904_ GG.KK.CPY.KY.LSV..Q.YIQNR....ET.L..SQR...TDRQV.I........Q..L--AAF81904 Branchiostoma-Hox14__AAF81905_ PV.PK.RPYSKY.LNE..N.YVQNQ.I..DK.LQ.SQK...T.RQV.I......I.Q.KL--AAF81905 Capitella-post1__EU196546_ NPKKK.KPYSKP.VSA..N.YSTST.ITKA..K.V.RE.D.T.RQ..I.Y....I.E.KI--EU196546 Lottia-post1__100031_ TL.KR.RPYSKF...E..R.YNGST...KS..W..SQLI..S.RQ..I......I.A.KI--JGI Lotgi1 protein id: 100031 Drosophila-AbdB__NM_080157_ SV.KK.KPYSKF.TLE.....LFNA...KQK.W...RN.Q.T.RQV.I........N.KN--NM_080157 Tribolium-AbdB__EEZ99247_ TV.KK.KPYSKF.TLE.....LFNA...KQK.W...RN...T.RQV.I........N.KN--EEZ99247 g_algea_scaff_4559_1.Contig10 TV.KK.KPYSKF.TLE.....LFNA...KQK.W...RN...T.RQV.I........S.KNAQR 15 : 71736 abdB g_algea_scaff_4559_3.Contig7 ------KPYSKF.TLE.....LFNS...KQK.W...RN...T.RQV.I........S.KSSQR 21 : 85120 abdB Capitella-post2__EU196545_ KQ.KK.KPY..Y.TMV..N..INNS.IT.QK.W.ISCK.H.S.RQV..........R.KL--EU196545 Lottia-post2__89720_ KG.KK.KPY..Y.TMV..N..LSSS.IT.QK.W.ISCK.Q.S.RQV..........R.KL--JGI Lotgi1 protein id: 89720 g_aosol_scaff_4355_1.Contig18 ---------------------EFNR.ITQR..I.I.HY.S.S.RK..M.....S..ANEKH-20 : 83145 hox? Drosophila-ftz__NM_058159_ DSK.T.QTY..Y.TLE.....HFNR.IT.R..IDI.N..S.S.RQ..I........S.KD--NM_058159 CDX-2__Mus_musculus_ TKDK..VVY.DH.RLE.....HFSR.ITIR.KS....T.G.S.RQV.I......A.ERKIKKK AAA19645 g_amvdv_scaff_4163_3.Contig12 TRDK..VVY.DQ.RLE.....HYNR.I.IR.KT....MVG.S.RQV.I......A.ER.NFRK 5 : 33811 cdx g_aildw_scaff_4772_1.Contig14 TKDK...VY.DH.RLE.....HYSR.ITIR.KT...TM.G.SDRQV.I......A.ERK.ARK 19 : 80739 cdx g_anovg_scaff_4508_1.Contig5 TKEK..VVY.DH.RLE......YSR.ITIR.KS....M.G.S.RQV.I......A.ERK.ARK 3 : 21395 cdx caudal__isoform_A__Drosophila_melanogaster_ TKDK..VVY.DF.RLE....YCTSR.ITIR.KS...QT.S.S.RQV.I......A.ERK.NKK NP_476954.1 Ftz__Tribolium_castaneum_ GNK.T.QTY..Y.TLE.....HFNK.LT.R..I.I.ES.R.T.RQ..I........A.KDTKF AF321227_1 Tribolium-ftz__AAC46491_ GNK.T.QTY..Y.TLE.....HFNK.LT.R..I.I.ES.R.T.RQ..I........A.KD--AAC46491 g_acrop_scaff_4310_1.Contig8 APK.T.QTY..Y.TLE.....HFNK.LT.R..I.I.HT.G.T.RQ..I........E.KENK15 : 71825 hox? g_awjsf_scaff_4464_1.Contig11 --K.T.QTY..Y.TLE.....HFNK.LT.R..I.I.HT.G.T.RQ..I....----------21 : 85120 hox? g_awjsf_scaff_4464_1.Contig15 APK.T.QTY..Y.TLE.....HFNR.LT.R..I.I.HT.G.T.RQ..I........A.KENKF 21 : 85120 hox? g_avhgw_scaff_4510_1.Contig7 --K.T.QTY..Y.TLE.....HFNR.LT.R..I.I.HS.G.T.RQ..I........A.KEMPK 18 : 79607 hox? g_alkdk_scaff_4677_1.Contig4 DQ..C.QTYSFY.TLA.....HFDP.LTKR..I.I.HT...K.KQ..I........W.KEKRS 28 : 94741 hox? g_ascty_scaff_4257_1.Contig45 DQ..S.QTYSFY.TLA.....HLDP.LTQR..I.I.HT...T.KQ.QI........W.KENKT 17 : 77601 hox? hox_6_Saccoglossus -Q..G.QTY..Y.TLE.....HFNR.LT.R..I.I.HT.G.T.RQ..I........W.KEQKABK00020 hox_7_Saccoglossus -KK.C.QTY..Y.TLE.....HYNR.LT.R..I..SHL.G.T.RQ..I........Y.KESKAAP79287 Capitella-lox5__EU196542_ EQK.T.QTY..Y.TLE.....HYNR.LT.R..I.I.H..Q.T.RQ..I........Y.KE--EU196542 g_anoef_scaff_4410_1.Contig27 PR..G.QSY..F.TLE.....HFNH.LT.R..I...HVVC.T.RQ..I........L.KEVRV 3 : 23600 hox? le:///Users/yjx/Dropbox/gigasciencereviews/gigascience_R2_s... 2 of 4 4/22/14, 12:20 PM

PAGE 3

name sequence accession position class '.' for match LG:coord. g_awjsf_scaff_4464_1.Contig6 ----G.Q.Y..F.TLE.....HYNH.LT.R..I.I.HVVC.T--------------------21 : 85120 hox? g_ajivj_scaff_4414_1.Contig2 ---.S.Q.Y..F.TLE.....HYNH.LT.R..I.I.HVVC.T--------------------15 : 71079 hox? g_anoef_scaff_4410_1.Contig17 --------------LE.....HYNH.LT.R..I.I.HVVC.T.RQ..I........L.KEIR. 3 : 23600 hox? Lottia-lox4__156931_ .R..G.QTYS.Y.TLE.....QFNH.LT.K..I.I.HT.C.T.RQ..I........M.KE--JGI Lotgi1 protein id: 156931 Capitella-lox4__EU196543_ .R..G.QTYS.Y.TLE.....QFNH.LT.K..I.I.H..C.T.RQ..I........L.KE--EU196543 g_awgsh_scaff_4551_1.Contig4 -R..G.QTY..F.TLE.....HFNH.LT.R..I.I.H..C.T.RQ..I........L.KEIR. 15 : 71699 hox? g_aqqvq_scaff_4781_1.Contig5 ------------------...HFNH.LT.R..I.I.H..C.T.RQ..I........L.KEIR. 21 : 85040 hox? g_asnox_scaff_4479_1.Contig10 ---------------------HFNH.LT.R..I.I.H..C.T.RQ..I........L.KEIR. 15 : 71793 hox? g_asnox_scaff_4479_1.Contig18 ---------------------HFNH.LT.R..I.I.H..C.T.RQ..I........L.KEIR. 15 : 71793 hox? Drosophila-abdA__NM_057345_ PR..G.QTY..F.TLE.....HFNH.LT.R..I.I.H..C.T.RQ..I........L.KE--NM_057345 Tribolium-abdA__AAB70263_ PR..G.QTY..F.TLE.....HFNH.LT.R..I.I.H..C.T.RQ..I........L.KE--AAB70263 g_aqqvq_scaff_4781_1.Contig10 --------YP.Y.TLE.....HTNH.LT.R..I.M.H..C.T.RQ..I........L.KEIQ. 21 : 85040 ubx g_asnox_scaff_4479_1.Contig17 -------------------------.LT.R..I.M.H..C.T.RQ..I........L.KEIQ. 15 : 71793 ubx g_asnox_scaff_4479_1.Contig21 LR..G.QTY..Y.TLE.....HTNH.LT.R..I.M.H..C.---------------------15 : 71793 ubx g_aityy_scaff_4289_1.Contig2 LR..G.QTY..Y.TLE.....HTNH.LT.R..I.M.H..C.T.RQ..I......--------10 : 54444 ubx Drosophila-Ubx__NM_080500_ LR..G.QTY..Y.TLE.....HTNH.LT.R..I.M.H..C.T.RQ..I........L.KE--NM_080500 Tribolium-Ubx__EEZ99249_ LR..G.QTY..Y.TLE.....HTNH.LT.R..I.M.H..C.T.RQ..I........L.KE--EEZ99249 g_asnox_scaff_4479_1.Contig28 LR..G.QTY..Y.TLE.....HTNH.LT.R..I.M.H..C.T.RQ..I........L.KEIQ. 15 : 71793 ubx Capitella-lox2__EU196544_ .R..G.QTY..Y.TLE.....KFNR.LT.R..I..SHM.C.T.RQ..I........E.KE--EU196544 Lottia-lox2__185752_ .R..G.QTY..F.TLE.....KFNR.LT.R..I..SHM.C.T.RQ..I........E.KE--JGI Lotgi1 protein id: 185752 Branchiostoma-Hox8__ABX39492_ ER..G.QTYS.Y.TLE.....HFNK.LT.R..I.I.H..G.T.RQ..I........L.KE--ABX39492 Branchiostoma-Hox7__ABX39491_ ERK.G.QTY..Y.TLE.....HFNK.LT.R..I.I.H..C.T.RQ..I........W.KE--ABX39491 Lottia-lox5__89598_ EQK.T.QTY..Y.TLE.....HFNR.LT.R..I.V.HM.G.T.RQ..I........W.KE--JGI Lotgi1 protein id: 89598 Branchiostoma-Hox6__ABX39490_ EKK.G.QTY..Y.TLE.....HFNK.LT.K..I.I.HL.G.T.RQ..I........W.KE--ABX39490 g_aubsu_scaff_4356_1.Contig13 ERK.G.QTY..Y.TLE.....HFNR.LT.R..I.I.H..C.T.RQ..I........W.KENK. 15 : 71793 hox? g_aubsu_scaff_4356_1.Contig10 ERK.G.QTY..Y.TLE.....HFNR.LT.R..I.I.H..C.T.RQ..I........W.KENK. 15 : 71793 hox? Drosophila-Antp__NM_079525_ ERK.G.QTY..Y.TLE.....HFNR.LT.R..I.I.H..C.T.RQ..I........W.KE--NM_079525 Tribolium-Antp__EEZ99250_ ERK.G.QTY..Y.TLE.....HFNR.LT.R..I.I.H..C.T.RQ..I........W.KE--EEZ99250 Capitella-Antp__EU196547_ ERK.G.QTY..Y.TLE.....HFNR.LT.R..I.I.H..C.T.RQ..I........W.KE--EU196547 Lottia-Antp__177860_ NRK.G.QTY..Y.TLE.....HFNR.LT.R..I.I.H..C.T.RQ..I........W.KE--JGI Lotgi1 protein id: 177860 Hox8_Metacrinus -KK.G.QTY..Y.TLE.....HFNR.LT.R..I.I.Q.VC.S.RQ..I........A.KETSBAF43726 g_atxvn_scaff_4590_1.Contig3 --------------------------LT.R..I.I.H..C.T.RQ..I........W.QENK. 5 : 33116 hox? g_awjsf_scaff_4464_1.Contig1 --K.G.QTY..Y.TLE.....HFNR.LT.R..I.I.--------------------------21 : 85120 hox? g_awjsf_scaff_4464_1.Contig8 ERK.G.QTY..Y.TLE.....HFNR.LT.R..I.I.H..C.T.R------------------21 : 85120 hox? g_ascty_scaff_4257_1.Contig27 ------------.TLE.....HFNR.LT.R..I.I.H..C.T.RQ..I--------------17 : 77601 hox? g_alkdk_scaff_4677_1.Contig1 -------TY..Y.TLE.....HFNR.LT.R..I.I.H..C.T--------------------28 : 94741 hox? hox_5_Saccoglossus -AK.S...Y..Y.TLE.....HFNR.LT.R..I.I.H..G.S.RQ..I........W.KEHNNP_001158410 Branchiostoma-Hox5__ABX39489_ DNK.T...Y..Y.TLE.....HFNR.LT.R..I.I.H..C.T.RQ..I........W.KE--ABX39489 Scr__Tribolium_castaneum_ ETK.Q..SY..Y.TLE.....HFNR.LT.R..I.I.H..C.T.RQ..I........W.KEHKM AF321227_2 Drosophila-Scr__X14475_ ETK.Q..SY..Y.TLE.....HFNR.LT.R..I.I.H..C.T.RQ..I........W.KE--X14475 Tribolium-Scr__EEZ99252_ ETK.Q..SY..Y.TLE.....HFNR.LT.R..I.I.H..C.T.RQ..I........W.KE--EEZ99252 g_anoef_scaff_4410_1.Contig19 ------------------...HFNR.LT.R..I.I.H..C.S.RQ..I........W.KEHKM 3 : 23600 hox? g_asnox_scaff_4479_1.Contig12 -------SY..Y.TLE.....HFNR.LT.R..I.I.H..C.S.RQ..I--------------15 : 71793 hox? g_awjsf_scaff_4464_1.Contig7 ETK.Q..SY..Y.TLE.....HFNR.LT.R..I.I.H..C.S.RQ..I........W.KEH-21 : 85120 hox? g_anbio_scaff_4694_1.Contig9 ETK.Q..SY..Y.TLE.....HFNR.LT.R..I.I.H..C.S.RQ..I........W.KEHKM 15 : 71825 hox? le:///Users/yjx/Dropbox/gigasciencereviews/gigascience_R2_s... 3 of 4 4/22/14, 12:20 PM

PAGE 4

name sequence accession position class '.' for match LG:coord. g_anbio_scaff_4694_1.Contig10 ETK.Q..SY..Y.TLE.....HFNR.LT.R..I.I.H..C.S.RQ..I........W.KEHKM 15 : 71825 hox? g_awjsf_scaff_4464_1.Contig5 --K.Q..SY..Y.TLE.....HFNR.LT.R..I.I.H..C.S--------------------21 : 85120 hox? Lottia-Scr__86927_ DSK.S..SY..H.TLE.....HYNK.LT.R..I.I.H....T.RQ..I........W.KD--JGI Lotgi1 protein id: 86927 Capitella-Scr__EU196541_ DNK.T..SY..H.TLE.....HFNR.LT.R..I.I.HS...T.RQ..I........W.KE--EU196541 pancreas_duodenum___1__Mus_musculus_ ENK.T...Y..A.LLE.....LFNK.I.....V...VM...T.RH..I........W.KEEDK NP_032840 Branchiostoma-Hox2__AB028207_ SS..L..V..NT.LLE.....HYNK..CK...K.I.SY.D.N.RQV.I.......RQ..R--AB028207 g_axtpu_scaff_4896_1.Contig3 ---SG..N..TK.L....N..LFNK.LT.K..I.I...IQ.N..QV.I........Q.KRMKE 9 : 53730 labial g_axtpu_scaff_4896_1.Contig4 ---SG..N..TK.L.G..N..LFNK.LT.K..I.I...IQ.N..QV.---------------9 : 53730 labial Drosophila-labial__NM_057265_ -NNSG..N..NK.LTE.....HFNR.LT.A..I.I.NT.Q.N..QV.I........Q.KRV-NM_057265 Tribolium-labial__AAF64149_ CLNTG..N..NK.LTE.....HFNK.LT.A..I.I.S..Q.N..QV.I........Q.KR--AAF64149 g_axtpu_scaff_4896_1.Contig2 --------------------------LT.A..I.I.T..Q.N..QV.I........Q.KRMKE 9 : 53730 labial g_ajwnd_scaff_4604_1.Contig5 -VGSG..N..TK.LTE.....HFNK.LT.A..I.I.T..Q.N..QV.I........Q.KRMKE 15 : 71825 labial g_aavsb_scaff_4656_1.Contig15 ----G..N..TK.LTE.....HFNK.LT.A..I.I.N..Q.N..QV.I........Q.KRMKE 18 : 79564 labial g_arkho_scaff_4547_1.Contig2 ----G..N..TK.LTE.....HFNK.LT.A..I.I.NV.Q.N..QV.I........Q.----1 : 5456 labial g_aavsb_scaff_4656_1.Contig7 ------------------------------..I.I.NV.Q.N..QV.I........Q.KRMKE 18 : 79564 labial Branchiostoma-Hox1__AB028206_ GPNNG..N..TK.LTE.....HYNK.LT.A..V.I......N..QV.I........Q.KR--AB028206 Capitella-labial__EU196537_ -PNMG..N..NK.LTE.....HFNK.LT.A..I.I..S.G.N..QV.I........Q.KRL-EU196537 Lottia-labial__100648_ -PNSG..N..NK.LTE.....HFNK.LT.A..I.I..S.G.N..QV.I........Q.KR--JGI Lotgi1 protein id: 100648 Lottia-pb__110623_ GS..L...Y.NT.LLE.....HFNK.LC....I.I..S.D.T.RQV..........Y...--JGI Lotgi1 protein id: 110623 Drosophila-pb__NM_057322_ LP..L...Y.NT.LLE.....HFNK.LC....I.I..S.D.T.RQV..........H...--NM_057322 Tribolium-pb__EEZ99256_ LP..L...Y.NT.LLE.....HFNK.LC....I.I..S.D.T.RQV..........H...--EEZ99256 g_adlvn_scaff_4304_2.Contig14 MP..L...Y.NT.LLE.....HFNK.LC....I.I..S.D.S.RQV..........H...TMM 15 : 71870 ? g_agkld_scaff_4527_1.Contig4 MP..L...Y.NT.LLE.....HFNK.LC....I.I..S.D.S.RQV..........H...TVM 3 : 19288 ? g_apppz_scaff_4373_1.Contig11 MP..L...Y.NT.LLE.....HFNK.LC....I.I..S.D.S.RQV..........H...TTM 21 : 85040 ? g_adlvn_scaff_4304_2.Contig9 MP..L...Y.NT.LLE.....HFNK.LC....------------------------------15 : 71870 ? Capitella-pb__EU196538_ HP..L...Y.NT.LLE.....HFNK.LC....I.I..S.D.T.RQV..........F...--EU196538 Tribolium-zen__NP_001036813_ AGK.A...Y.SA.LVE..R..HHGK.L.....IQI.EN...S.RQ..I........H.KE--NP_001036813 g_adhdu_scaff_4669_1.Contig2 PTK.V...Y.SA.LVE.....HFNR.LC....I.M.NL...T.RQ..I........Y.K---21 : 85075 hox3 g_amdsu_scaff_4454_1.Contig21 --K.A...Y.SA.LVE.....HFNR.LC....V.M.NL.S.T.RQ..I........Y.KEQKN 15 : 71825 hox3 Capitella-Hox3__EU196539_ PSK.A...Y.SA.LVE.....HFNR.LC....I.M..L...T.RQ..I........Y.KD--EU196539 Lottia-Hox3__56601_ PAK.A...Y.SA.LVE.....HFNR.LC....I.M..L...S.RQ-----------------JGI Lotgi1 protein id: 56601 g_amdsu_scaff_4454_1.Contig17 --K.P...Y.TA.LTE.....HFTR.LC....V.M.GL.H.T.RQ..I.......-------15 : 71825 hox3 Branchiostoma-Hox3__X68045_ AGK.A...Y.SA.LVE.....HFNR.LC....V.M..M...T.RQ..I........Y.KE--X68045 Drosophila-zen__AAF54087_ KLK.S.....SV.LVE..N..KSNM.LY.T..I.I.QR.S.C.RQV.I........F.KD--AAF54087 g_aheja_scaff_4145_1.Contig5 KTK.T..S.NGV.LLE.....TNNM.L..L..I.I.NY...S.KQV.I......V.F.KEG-23 : 89444 gsx g_atxvn_scaff_4590_1.Contig4 SSKKI..VY..S.LLE..R..AANM.LT.L..I.I.TY.S.S.KQV.I......V.Y.KE--5 : 33116 gsx g_aneyu_scaff_4415_1.Contig10 SSK.I.....ST.LLE..R..SANM.L..L..I.I.TY...S.KQV.I......V.Y.KE--6 : 37542 gsx Gsx1___Mus_musculus_ SSK.M.....ST.LLE..R..ASNM.L..L..I.I.TY...S.KQV.I......V.H.KEGKG AAI37770.1 g_aucja_scaff_4557_1.Contig17 --K.I.....SA.LLE..R..STNM.L..I..I.ISKY...S.KQV.I......V.H.KEG-19 : 80662 gsx le:///Users/yjx/Dropbox/gigasciencereviews/gigascience_R2_s... 4 of 4 4/22/14, 12:20 PM