微生物生态：科学问题与实验设计

微生物生态：科学问题与实验设计微生物生态：科学问题与实验设计微生物生态：科学问题与实验设计微生物生态：科学问题与实验设计微生物生态：科学问题与实验设计1致谢周集中李筱婧陈永健束浩月邝嘉良花正双致谢周集中2InstituteforEnvironmentalGenomics,DepartmentofMicrobiologyandPlantBiology,UniversityofOklahoma,Norman,OK73019MicrobialEcologyinGenomicsEra:ChallengesandOpportunitiesZhongshanUniversity,Guangzhou,November5,2015.Jizhong(Joe)Zhoujzhou@;405-325-6073SchoolofEnvironment,TsinghuaUniversity,BeijingEarthScienceDivision,LawrenceBerkeleyLaboratory,Berkeley,CAInstituteforEnvironmentalGe3共同关心的问题采样量（体积、重量）测序深度多少样品是否重复是否混样16S测序策略宏基因组与GeoChip16S与PICRUSt分析数据分析回答这些问题的根本：科学问题共同关心的问题采样量（体积、重量）4专业：生态学

答辩人：李筱婧

导师：黄立南教授

中山大学硕士学位论文答辩采样量与估算方法

对细菌物种丰富度估算的影响专业：生态学

答辩人：李筱婧

导师：黄立南5研究背景稳健的微生物物种丰富度评价对于研究微生物多样性及微生物群落的组成与功能的是必不可少的1微生物个体数目巨大，多样性极高，占据了地球生物多样性的绝大部分(Whitmanetal.,1998;Torsviketal.,2002)；2微生物的物种丰富度与组成一定程度上决定微生物群落功能(Belletal.,2005)，并在地球生物化学循环中扮演核心角色(Prosseretal.,2007)；3不同丰度水平的物种对细菌群落水平的多样性有着不同程度的贡献，且在微生物群落中有着不同的角色和功能(Shade&Handelsman,2011;Campbelletal.,2011)。研究背景稳健的微生物物种丰富度评价对于研究微生物多样性及微生6材料与方法采样

理化Miseq

测序多样性分析细胞

密度序列

处理1234567提取DNAqPCR

验证采样样品准备数据分析材料与方法采样

理化Miseq

测序多样性分析细胞

密度序列7采样采样地：广东大宝山矿区多金属矿

采样时间：2014年9月25-26日采样水量：1ml、5ml、10ml、50ml、100ml、500ml、1L、5L和10L等9个水平（每个水平10个重复）采样操作：1原位收集微生物细胞2原位测定温度、pH值、溶氧值和Eh值3保存适量水样用于细胞计数和理化测定采样采样地：广东大宝山矿区多金属矿

采样时间：2014年98采样量的影响采样量的影响9整体上单个样品所检测到的OTUs数目与采样量成正比，但采样量间OTUs数目无显著性差异；当采样量超过100ml，OTUs数目逐渐稳定在同一水平,主要类群分布相似1图3-1不同采样量间AMD细菌群落的OTUs分布图3-2不同采样量间AMD细菌群落的主要类群分布SampleVolume采样量的影响(测序量100000)

物种丰富度与组成整体上单个样品所检测到的OTUs数目与采样量成正比，但采样量10测序量的影响测序量的影响11图3-4不同测序量间AMD细菌群落的主要类群分布测序

深度100010000100000门10

major->92.1%7.9%unassigned13

major->89.4%

9.1%unassigned23

major->90.3%

9.6%unassigned纲152334目254268科2651115属19

71%unassigned53

76%unassigned304

76%unassignedOTUs151~254537~9021632~2432测序量的影响

物种丰富度与组成整体上单个样品所检测到的物种门类和丰富度与测序量成正比，但测序量间主要类群分布无显著性差异。1表3-1不同测序量间AMD细菌群落的物种组成分布图3-4不同测序量间AMD细菌群落的主要类群分布测序

深度12测序量的影响（500ml样品）

常见种和稀有种的覆盖度随着测序量的上升，新增的常见种逐渐减少，而稀有种的增幅不一；测序量超过10000条即可检测到50%的常见种，而测序量需超过50000才可检测到50%的稀有种。3图3-6AMD细菌群落中常见种和稀有种的数目

（柱图表示新增数目，折线图表示累积数目）98%80%测序量的影响（500ml样品）

常见种和稀有种的覆盖度随着测1397%cutoffThreerunsofMiSeqOTUOverlapsamongThreeTagswithMiSeq54soilsamplesSameDNAsfromeachsampleamplifiedandsequencedbythreetags>10,000sequencespertechnicalreplicate<15%ofoverlapforwasobservedforidenticalcommunitiesThisissimilartopyrosequencing,indicatingthatthisartifactissequencing-technologydependent

(A)

(C)(B)(D)12.9±0.9%16.03%13%16.03%13%16.03%13%11.2±0.9%17.4%12.2%17.4%12.2%17.4%12.2%13.7±0.9%15.07%13.7%15.07%13.7%15.07%13.7%11.5±0.6%16.7%12.8%16.7%12.8%16.7%12.8%97%cutoffOTUOverlapsamongT14EffectsofSequencingEffortonOTUOverlaps:neverreach100%UclustUPARSE

Deepsequencingto16MreadspersampleOTUoverlapincreaseswithsequencingdepth,butwithsaturationat60%30Kpersamplecouldbereasonablesampleeffortspersample15EffectsofSequencingEfforto估算方法的影响估算方法的影响16模型LognormalPareto

(powerlaw)Exponential参数

*KStest

1000

01998BESTLognormal

PowerLaw

Exponential图3-7AMD细菌群落物种丰度分布模型拟合比较估算方法的影响（500ml样品，测序量50000）

参数估算利用三种常用参数模型拟合并经KS检验后，Exponential模型的拟合效果比Lognormal和PowerLaw更佳。1表3-2不同拟合模型间参数和KS检验结果模型LognormalPareto

(powerlaw)E17结论进行采样设计时，采样量设置为100-500ml，每个样品的测序量为30000，可以获得比较稳定的物种丰富度观测值和物种组成；123进行数据分析时，应考虑综合考虑序列数据的取舍标准和估算方法的选择，尤其是参数模型的选取与比较，需要综合考虑录入模型的数据范围、验证方法以及拟合优度评价体系；本研究所得到的物种丰富度估算虽然可以用于研究细菌群落组成和丰度分布，但对于细菌种子库的研究，尤其是稀有种的估算，还需要结合时间或空间系列的样品才能更好地估算稀有种的丰富度。结论进行采样设计时，采样量设置为100-500ml，每18CEE结论不同的目的选择不同的测序深度绝对多样性的估计：测序深度30000及以上研究方法的探索相对多样性的估计：测序深度10000足够绝大多数研究水体采样量100-500ml（16S）正确选择估算参数更重要CEE结论不同的目的选择不同的测序深度19UncoveringthesoilmicrobialdiversityinasubtropicalforestReporter:YongjianChenSupervisor:Prof.WenshengShuResearchinterests:communityecologybiogeographyecophylogeneticsUncoveringthesoilmicrobial20Soilmicrobialdiversityinforest/Heishiding(HSD)plotHeishidingNatureReserve(111

49'09''~111

55'01''East,23

25'15''~23

30'02''North),GuangdongprovinceofChinaSubtropicalevergreenbroad-leavedforestArea:50ha(1000×500m)all218838free-standingplantswithdiameter≥1cmatbreastheight(DBH),representing237speciesweretaggedandmappedspatiallyCTFSnetworkSoilmicrobialdiversityinfo21Soilsamplingandanalyses30m30m1708soilsamplesfocalpointMeasuresoilchemicalproperties:moisture,pH,electricconductivity,organiccarbon,nitrogen,P,Ca,Mg,AlDNAextractionPCRamplificationV4regionof16srDNA(bacteria)ITS2(fungi)250bppair-endsequencingonIlluminaMiSeqgrids:30×30mSoilsamplingandanalyses30m221.1SoilmicrobialcommunitycompositionBacterialassemblagesarephylum-rich(9phylawithrelativeabundance>1%)butvaryslightlyamongsamplesFungalassemblagesarephylum-poor(3phylaconsistedmostofallfungalsequences)butvaryremarkablyamongsamples1.1Soilmicrobialcommunityc231.4Species-accumulationcurvesNumberofOTUstendedtobeasymptotictotheupperboundvalueasnumberofsamplesincrease,implyingthatfewnewlyobservedspeciesweredetectedtowardstheendofoursampling1.4Species-accumulationcurve24Tedersooetal.,Science201440soilcoresfromnaturalcommunitiesineachof365sitesacrosstheworldamixtureofsixITS3primersandITS4ngsfivehalf-plates454MapsofglobalsamplingTedersooetal.,Science2014425Sampling32forests192sites3replications/siteGradientsGeographyClimateVegetationElevationSoilEnv.Scales10m~3,000kmSampling32forests263.Conclusions

Themostlikelylower-boundestimatedOTUsnumberis39,244(95%confidenceinterval(CI)=39,221–39,274)and14,223(95%CI=14,221–14,225)forbacteriaandfungi,respectively.Niche-basedenvironmentalcontrolprocessesdominantlygovernsoilbacterialcommunityassembly,whereasdispersal-basedspatialprocessesappeartobethemaindriverofsoilfungalcommunityassembly.Similarbelowgroundsoilmicrobialcommunitycompositionisaccompaniedbysimilarabovegroundplantcommunitycomposition.3.ConclusionsThemostlikely27CEE结论对格局的研究而言，样品多多益善大样地研究，500个样品是基本数量样品数量的重要性，远高于测序深度的重要性对于最复杂的土壤样品，3000-5000有效序列足够对于一个和几个样品的情况，3000-5000有效序列也足够，但一定要结合宏基因组/宏转录组/宏蛋白组/宏代谢组/SIP等/单细胞基因组等深度研究与个性化分析CEE结论对格局的研究而言，样品多多益善2816S测序策略V3（150）；V4（320）；V3+V4（480）；V4+V5（400-450）；V6建议：V4；V4+V5不建议：V3；V6；V3+V4理由：测序质量/数据量/分析精度/文献可比性的平衡16S测序策略V3（150）；V4（320）；V3+V4（429宏基因组与GeoChip宏基因组与GeoChip30Whatarethefunctional

activitiesofamicrobialcommunity?Whataretherelationshipsbetweenmicrobialcommunityfunctional

structureandecosystemfunctioning?Howdoesmicrobialcommunityfunctionalstructurerespondtoenvironmentalchangessuchasclimatewarming?FromstructuretofunctionsWhatarethefunctionalactivi31OpenformatdetectionCouldnotassurethesamegenes/proteins/organismscanbecomparedacrossdifferentsamples.TheresultscouldbeexpectedandthusareopenHighthroughputSequencing454sequencing,Illumina,PacBioProteomics,MetabolomicsFunctionalmetagenomicsSinglecellgenomicsClosedformatEnsurethatthesamegenes/proteins/organismscanbecomparedacrossdifferentsamples.Theresultscanbeexpected,andthusareclose.BioLogPlatesPhyloChip:16SgenesGeoChip:functionalgenesI.HighThroughputOmicTechnologiesOpenformatdetectionI.HighT3297%cutoff1runofpyrosequencing11.7±3.3%13.3±3.3%7.4±2.1%9.1±2.5%24.3%6.0%25.1%5.8%24.3%6.0%25.1%5.8%24.3%6.0%25.1%5.8%20.1%8.8%20.3%9.1%20.1%8.8%20.3%9.1%20.1%8.8%20.3%9.1%(A)(C)(B)(D)ForwardprimerSingletonsnotremovedForwardprimerSingletonsremovedReverseprimerSingletonsnotremovedReverseprimerSingletonsremovedOTUOverlapsamongThreeTagswithpyrosequencing24samples,warmingvsnonwarming60tags<15%ofoverlapwasobservedforidenticalcommunitiesTechnicalvariationsarebigZhouetal.2011.ReproducibilityandQuantitationofAmpliconSequencing-BasedDetection.ISMEJ.5:1303-1313.97%cutoff11.7±3.3%13.3±3.3%7.33Technicalreproducibilitymeasuredbysimilarityamongtechnicalreplicates6forestsitesfromNorthAmerica(PanamatoOregon),3soilsamples/site,3technicalreplicates/soilsample,altogether,54datasets.AlltechnicalreplicateswereanalyzedwithGeoChip(180Kprobes),16Ssequencing(>80KReads),andshotgunsequencing(>10Gbpersample).Sorensensimilarity:Qualitative;Bray-Curtis:QuantitativeTechnicalreproducibilitymeas34Replicates!Replicates!Replicates!IncreasingsamplingeffortsAmelioratingthepotentialproblemoflowreproducibilityZhou,etal.2015.High-ThroughputMetagenomicTechnologiesforComplexMicrobialCommunityAnalysis:OpenandClosedFormats.mBio6(1):e02288-1435Replicates!AmelioratingthepCEE评论对于任何实验研究，重复都是必须的对于生态学调查，重复一般是以环境梯度/样品的空间分布替代的永远不要混样CEE评论对于任何实验研究，重复都是必须的36Microarrays:

Glassslidesorothersolidsurfacecontainingthousandsofgenesarrayedbyautomatedequipment.FGAscontainprobesfromthegenesinvolvedinvariousgeochemical,ecologicalandenvironmentalprocesses.C,N,S,PcylcingsOrganiccontaminantdegradationMetalresistanceandreductionAntibioticresistanceTypicalformat:50meroligonucleotidearraysUsefulforstudyingmicrobialcommunitiesFunctionalgenediversityandactivityLimitedphylogeneticdiversity.GeoChiporFunctionalGeneArrays

(FGAs)Microarrays:Glassslidesoro37GeoChip:AhighthroughputtoolforlinkingcommunitystructuretofunctionsHe,Z,TJGentry,CWSchadt,LWu,JLiebich,SCChong,ZHuang,WWu,BGu,PJardine,CCriddle,andJ.Zhou.2007.GeoChip:acomprehensivemicroarrayforinvestigatingbiogeochemical,ecologicalandenvironmentalprocesses.TheISMEJ.1:67-77.Highlightedby:

ApressreleasebyNaturePressOfficeReportedbymanyNewspapers

NationalEcologyObservatoryNetworks(NEON),RoadmapNationalAcademyofSciences,MetagenomicsreportR&D100,amongmostoutstanding100technologicalinnovationsandbreakthroughin2009GeoChip:Ahighthroughputtoo38R&D100,amongmostoutstanding100technologicalinnovationsandbreakthroughin2009R&D100AwardR&D100,amongmostoutstand39GeoChipRoadmap20022005200720132402

Genes24,243Probes150

GeneFamilies28,000Probes292

GeneFamilies83,992Probes549

GeneFamilies2010268,059Probes2,433

GeneFamiliesThemostcomprehensivehighthroughputmetagenomictechnologyforcharacterizingmicrobialcommunityfunctionalstructureandactivities.GeoChipRoadmap20022005200720140SummaryofGeoChip5.0probeinformationbyfunctionalgenecategoryGenecategoryNo.ofgenesincategoryNo.ofprobesdesignedNo.ofcoveredsequencesCarboncycling1714499787764Nitrogen401150324631Sulphur28850312657Phosphorus637837141Antibioticresistance352293459141MetalResistance14556552134681OrganicRemediation3493079659064Stress1233411987270Algalvirus1915382981Bacteriaphage4110952083EnergyProcess1112561647Environtoxins36191794Fungifunction109860411457Pigments3632294910Protists8329944017Soilbenefit50955914424Soilbornepathogeirulence898879222532Virus4438629462othercategory511184719874Total2,433268,059570,042SummaryofGeoChip5.0probei41Detectingfunctions:Geochemicalprocesses,ecologicalprocessesBroaddetection:~2500functionalgenecategoriesofimportancetoecology,geochemistryandenvironmentalsciencesHigherresolution:Species-strainlevelresolutionQuantitative:noPCRisinvolvedLowcost:<

$0.2gene/sampleSpeed:VeryquickMainadvantagesofGeoChip

(Comparedtootherapproaches,16S-based454sequencing,PhyloChip)Detectingfunctions:Geochemic42GeoChipFish-outWorkflowGeoChipFish-outWorkflow43KeydifferencesbetweenopenandcloseformatdetectionOpenformatCloseformatSensitivitytorandomsamplingerrorsHighLowEffectsbydominantorganismsYesNoFindingnewthingsYesNo/YesSensitivitytocontaminatedDNAsYesNoDataanalysisVeryslowrightnowRapidZhou,etal.2015.High-ThroughputMetagenomicTechnologiesforComplexMicrobialCommunityAnalysis:OpenandClosedFormats.mBio6(1):e02288-14Keydifferencesbetweenopena44EnvironmentalremediationBiodiversityWangetal.2009.PNAS,106:4840-4845Zhouetal.2008.PNAS,105:7768-7773zvaluesarelessthan0.1ClimatechangeZhouetal.,2011.NatureClimateChange,2:106-110Chanetal.2013.PNAS,110:8990-8995Hazenetal.2010.Science,330:204-208

Zhouetal.2014.PNAS,111:E836-E845

EcologicalTheoryWangetal.,2014.NatureComm.DOI:10.1038/ncomms5799EnvironmentalremediationBiodi45CEE结论复杂样品，推荐Geochip实验研究的处理与重复样品，推荐Geochip旨在新发现（新基因/新的代谢途径/进化/基因水平转移/基因突变等）的研究：高通量测序CEE结论复杂样品，推荐Geochip46PICRUSt分析：基于16S的微生物宏基因组预测Reporter:ShuHao-yueLaboratoryofPollutionandRestorationEcologyPICRUSt分析：基于16S的微生物宏基因组预测Repor47微生物生态：科学问题与实验设计48AcentralquestioninbiologyishowbiodiversityinfluencesecosystemfunctioningItremainsunclearhowmicrobialcommunitycompositionisrelatedtoecosystemfunctioningThekeytoaddressingthisissueistherelationshipbetweenphylogenyandfunctionaltraitsbiodiversityIntroductionecosystemfunctioningAcentralquestioninbiology49Hypothesis:

functionaltraitsbasedoncomplexgeneticsystemsintegraltotheorganismevolveslowlyandwillbemorephylogeneticallyconserved(traitA).simpletraitsconsistingoffewinteractingproteinswillbemoredispersedorcompletelyrandomlydistributed(traitB,C)Hypothesis:5093%(83out89)ofthetraitswerenonrandomlydistributed93%(83out89)ofthetraits51微生物生态：科学问题与实验设计52PICRUStalgorithmPICRUStalgorithm53PICRUStrecapitulatesHMPmetagenomes530(HMP)samples(both16SrRNA&shotgunmetagenomesequencing),>700draftandfinishedreferencegenomes.Spearmanr=0.82,P<0.0