[PDF] Single cell RNA sequencing Sequencing cost becomes the bo

View - Download Single cell RNA sequencing

Previous PDF

Next PDF

SinglecellRNAsequencing

asa.bjorklund@scilifelab.se

Outline• Whysinglecelltranscriptomics?

• Experimentalsetup• ComputaAonalanalysis• ExamplesofscRNA-seqexperiments (Sandberg,NatureMethods2014) Whysingle-celltranscriptomics?• UnderstandingheterogeneousAssues

• IdenAficaAonandanalysisofrarecelltypes• ChangesincellularcomposiAon• DissecAonoftemporalchanges• ExampleofapplicaAons:

- DifferenAaAontrajectories- Cancerheterogeneity- NeuralcellclassificaAon- Embryonicdevelopment- Drugtreatmentresponse

Transcrip (Suteretal.Science2011)• BurstfrequencyandsizeiscorrelatedwithmRNAabundance • ManyTFshavelowmeanexpression(andlowburstfrequency)andwillonlybe detectedinafracAonofthecells

Experimentalsetup

Experimentalsetup

Single-cell transcriptomics protocols overviewKolodziejczyk et al, 2013

Experimentalsetup

Single-cell transcriptomics protocols overviewKolodziejczyk et al, 2013

Experimentalsetup

Single-cell transcriptomics protocols overviewKolodziejczyk et al, 2013 (Grunetal.Cell2015) Smallvolumeapproaches• VolumeseemtobeakeycomponentinthesereacAons - SmallervolumesgivebeeerdetecAonandreproducibility

• Smallervolumes=cheaperreagentcosts• Methodsforhighthroughput(1000ndsofcells)• Sequencingcostbecomestheboeleneckinstead-

ohenshallowsequencing

Droplet/microfluidicsapproaches

Chromium10XGenomics

• DropletbasedsystemforscRNAseqandgenomesequencing• 500-10,000singlecells• CellRangersofwareforanalysisofresults(GXYZhangetalNatCommunica4ons2017)

WafergenICELL8microwellsystem

• ICELL8chipwith9600wells• MulAsamplenanodispenser• CanuseFACSsorAng• ImagingstaAon• Sohwaretoselectcells->minimizenumberofdoublets• Possibletosaveimagesofcellswithafewdifferentcolors• ImplementedwithSTRTattheESCGplaqorm

ProblemscomparedtobulkRNA-seq• AmplificaAonbias

• Drop-outrates• TranscripAonalbursAng• Backgroundnoise• Biasduetocell-cycle,cellsizeandotherfactors

(Karchenkoetal.NatureMethods2014) (Shaleketal.Nature2013) • Cellularbarcodes - IntroducedatRTstepwithoneuniquesequencepercell- Enablespoolingofmanylibrariesintoonetubefor subsequentsteps • UMIs - Introducerandomsequencesatthebeginningofeach sequence - ReduceseffectofamplificaAonbiasbyremovingPCR duplicates • Implementedwithtag-basedmethodssuchasSTRT andCEL-seq (Islametal.NatureMethods2014)

Spike-inRNAs• AddiAonofexternalcontrols

• Usedtomodel: - technicalnoise/drop-outrates- starAngamountofRNAinthecell • ERCCspike-inmostwidelyused,consistsof48or96 mRNAsat17differentconcentraAons. • Importanttoaddequalamountstoeachcell, preferablyinthelysisbuffer.

Spike-inRNAs

(Vallejosetal.PLOSCompBiol2015) (Brenneckeetal.NatureMethods2013) • Recommendedtohaveatleast20-30cellsfromeachcelltype - Asamplewithaminorcelltypeat5%requiressequencingof

400cells.

- PreselecAngcellsmaybenecessary,butunbiasedcellpickingis preferred. - DependingonthesensiAvityofyourmethodyoumayneed more/lesscells • Tostudygeneexpressiononly,sequencingdepthdoes nothavetobedeep. - MulAplexingofhundredsofsamplesononelaneiscommon.- Fortag-basedmethodssequencingisohenmoreshallow. • Possibletohaveaconsultancysessionwithsomeoneat

NBISforexperimentaldesign.

WhichmethodshouldIuse?• Fulllength(SmartSeq2)vstag-based(CELseq/STRT)methods: - Trade-offbetweenthroughputandsensiAvity- UniquemolecularidenAfiers(UMI)implementaAonwith thetag-basedmethods • PracAcalissuessuchassorAngofcells NaRNA-seqservicesatESCG

Full-lengthQuan

FormatEppendorfTwin-tekC1microfluidicschip(Fluidigm)MicrowellchipChromiummicrofluidicschipCellnumber3843x969,600(~2,500)8x500-10,000InputFACS-sortedcellsCellsuspensionCellsuspensionCellsuspensionTranscriptcoverageFull-length5'5'3'Advantage• Flexibledelivery• SNPs,mutaAons• Nuclei• Imaging• CellselecAon• Unbiased• CellselecAon• 8samplesparallel• Nuclei• Highthroughput• 8samplesparallel• SamplepoolingLimita (FromKarolinaWallenborgatESCG)

Userfees

Smart-seq2384wellplateSTRT-C196cellchip(50-96cells)STRT-Wafergen9600wellschip(~2,500cells)10XGenomics1sample(~3,000cells)• ValidaAon• Smart-seq2library• Sequencing• (50bp,single-read• ValidaAon• STRT-C1library• Sequencing(50bpsingle-read)• ValidaAon• STRTlibrary(dualindex)• Sequencing(50bpsingle-read)• ValidaAon• Illuminalibrary• Sequencing(paired-end,dualindex)~40,500SEK~22,500SEK~45,000-50,000SEK~43,000SEK

scRNA-seqanalysisoverview

Mapping

GeneexpressionesAmate

QC:RemovelowQcellsRemovecontaminantsData:expressionprofilesRawdata-fastqfilesDefiningcelltypes/lineagesGenesignaturesVerificaAonexperiments• DimensionalityreducAon• Clusteringmethods• PseudoAmeassignmentSameasbulkRNAseq(withspike-ins)• DatanormalizaAon• GenesetselecAon• Batcheffectremoval• Removalofotherconfounders

QualityControl(QC)

uniquemap

9 cells removed, cutoff 8.3711

0 5 10 15 20 25
30

3.516.531.546.561.576.5

mismatch/indel

26 cells removed, cutoff 0.7076

0 10 20 30
40
50
60

0.230.570.911.251.59

exonmap

20 cells removed, cutoff 0.6052

0 10 20 30
40
50
60
70

0.0150.2250.4350.6450.855

3primemap

13 cells removed, cutoff 0.0856

0 50
100
150
200
250
300

0.02250.09250.16250.2325

normreads

16 cells removed, cutoff 100000.0000

0 10 20 30
40
50

50000175000036500005550000

gene_detection

14 cells removed, cutoff 0.0949

0 10 20 30
40
50
60

0.0350.1750.3150.4550.595

total 48 samples failed QC

Total samples: 722

• QCisacrucialstepinscRNA-seq-Anyexperimentwillhaveanumberoffailedlibraries! • OBS!Smallercelltypesgiveslowermappingratesandmoreprimer dimers. • Canlookat:

- MappingstaAsAcs(%uniquelymapping)- Mismatchrate- FracAonofexonmappingreads- 3'bias(degradedRNA)- mRNA-mappingreads- Numberofdetectedgenes- Spike-indetec • Dependingoncelltype,around500Kexonmappingreadssaturates Databias• OhenneedtododatatransformaAonbeforeclustering/PCA

- Normalizebyspike-inRNAs- Normalizebytotalcounts- LengthnormalizedRPKM/FPKM- Removecell-cycleeffects,sizebiasorsimilar(scLVM

package,SCDEpackage)

- RTefficiency/drop-outrate(SCDEpackage,scranpackage)- Technicalnoise(BASiCSpackage,GRM)- Batcheffectremoval(SVAComBatfuncAon,SCDEpackage)

Batchnormaliza -80-60-40-2002040 60
40
20 0 20 40
60

Raw RPKM values

PC1 PC2 ILC1 ILC2 ILC3 NK T74 T75 T86 -80-60-40-2002040 60
40
20 0 20 40
60

Raw RPKM values

PC1 PC2 T74 T75 T86 ILC1 ILC2 ILC3 NK -40-20020 40
20 0 20

Sva normalized RPKM values

PC1 PC2 ILC1 ILC2 ILC3 NK T74 T75 T86 -40-20020 40
20 0 20

Sva normalized RPKM values

PC1 PC2 T74 T75 T86 ILC1 ILC2 ILC3 NK (Stegleetal.NatRevGene4cs2015) Featureselec- GenesexpressedinXcells.- FilteroutgeneswithcorrelaAontofewothergenes- Priorknowledge/annotaAon- DEgenesfrombulkexperiments- TopPCAloadings

• Linearmethods: - PCA(principalcomponentanalysis)- ICA(independentcomponentanalysis)- MDS(mulAdimensionalscaling) • Non-linearmethods: - Non-linearPCA- t-SNE(t-distributedstochasAcneighborembedding)- Diffusionmaps- Networkbasedmethods • APCAisaverygoodstartingeyngtoknowyour t-SNEvsPCAdimensionalityreduc 20 0 20 PCA PC1 PC2 ILC1 ILC2 ILC3 NK T74 T75 T86 -40-2002040 40
20 0 20 t-SNE PC2 ILC1 ILC2 ILC3 NK T74 T75 T86

Moredimensionalityreduc (Moignardetal.NatureBiotech2015)

Iden • Clusteringbasedon - rpkms/counts-Euklideandistances- PairwisecorrelaAons- PCAorotherdimensionalityreducAonmethod • Methodofchoice:hierarchical,k-means,biclustering• Someprograms: - WGCNA- BackSPIN- Pagoda- DBscan • OBS!OutlierremovalasaniniAalstepmaybenecessary, (Fanetal.NatureMethods2016)

Pseudo (Trapnelletal.NatureBiotech2014)

Diffusionpseudo (HaghvedietalNatureMethods2016)

Detec • Canusenon-parametricmethodslikeSAMseq

Detec • AvailablesinglecellDEmethods:

- SingleCellAssay-developedforqPCRexperiments- Monoclepackage- SingleCellDifferenAalExpression-SCDE- Model-basedAnalysisofSingle-cellTranscriptomics-MAST- SAMstrt-extenAontoSAMseqwithspike-innormalizaAon- ManyotherrecentpublicaAons.......

• SomestudiesusePCAcontribuAon(loadings)orgene SAM SCA SCDE DESEQ

MONOCLE

MONOCLE

DESEQ SCDE SCA SAM

741557362755794

10496476192780755

399375689619362

631661375647557

14296313991049741

ILC3 overlap of DE genes

ComparisonofDEdetec

HighfalsediscoveryrateforDESeqandEdgeR

scRNA-seqanalysisoverview

Mapping

GeneexpressionesAmate

QC:RemovelowQcellsRemovecontaminantsData:expressionprofilesRawdata-fastqfilesDefiningcelltypes/lineagesGenesignaturesVerificaAonexperiments• DimensionalityreducAon• Clusteringmethods• PseudoAmeassignmentSameasbulkRNAseq(withspike-ins)• DatanormalizaAon• GenesetselecAon• Batcheffectremoval• Removalofotherconfounders

Addi

• Variantcalling• Copy-numbervariaAon• AlternaAvesplicing• AlternaAvesplicingandallelicexpressionrequires

fulllengthmethods.

- Butonlyworksforhighlyexpressedgeneswithgoodreadcoverage- MustbecarefultotakeintoconsideraAonthedrop-outrate,aunique

phenotypeLinkageAnalysis (Poironetal.BioRxiv2016)

EIEJCEU=IKPK=PAH

quotesdbs_dbs21.pdfusesText_27

[PDF] JO du 25 mars 2012 - SGG Algérie

[PDF] prise en charge des frais de deplacements professionnels - Cdg50

[PDF] frais de deplacement - CDG38

[PDF] MISE A JOUR DU 16 AOUT 2017 - Cdg59

[PDF] mémo TZR 2014 - Snes

[PDF] le livret du tzr - Académie de Toulouse

[PDF] Le Conseil d 'Administration du Centre Régional des - Crous de Lille

[PDF] DESTINATIONS, TEMPS DE TRANSIT, TARIFS 2017

[PDF] Référentiel économique du vigneron - VinOpole

[PDF] Volume 3 - Ministère de l 'Economie et des Finances

[PDF] Algerie - Reglement de la Banque d 'Algerie n°07-01 relatif aux

[PDF] 02 Prise en charge des frais de déplacement- MAJ 13012014

[PDF] bareme d honoraires transaction - Square Habitat

[PDF] Frais de restructuration - Traitement dans les comptes annuels - CBN

[PDF] Frais de scolarité des étudiant-e-s étrangers du premier cycle