The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

52
THE RNASEQ ANALYSIS PIPELINE Alicia Oshlack Murdoch Childrens Research Ins5tute

description

High-throughput sequencing of the transcriptome leads to the production of millions of short reads which need to be analysed in order to make biological inference. RNA-seq data are complex and the analysis involves a series of steps for which research is ongoing. However the choice of analysis methodology can have a major impact on the findings of an experiment. In this talk I will outline the major steps involved in analysing an RNA-seq dataset for the purpose of detecting differential expression. I will discuss and compare the available strategies for each stage of the pipeline. I will also point out several areas in RNA-seq data analysis which require further research in order for the full potential of the technology to be Realised.

Transcript of The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Page 1: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

THE  RNA-­‐SEQ  ANALYSIS  PIPELINE  

Alicia  Oshlack  Murdoch  Childrens  Research  Ins5tute  

Page 2: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Genes  and  transcripts  

Gene  

transcript  

Page 3: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

New genomic technologies are driving the science

Page 4: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Two  ways  to  look  at  sequencing  data  

Sequence  of(mapped)  read  • genome  sequencing    • variant  detec5on  • Muta5on  detec5on  • genomic  rearrangements  • Bisulfite-­‐seq  (methyla5on)  • RNA  edi5ng  etc.  

Posi5on  of  mapped  read  • RNA-­‐seq  • ChIP-­‐seq    • MeDIP-­‐seq  for  DNA  methyla5on  etc.  

   

4  

Page 5: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Two  ways  to  look  at  RNA-­‐seq  data  

Sequence  of(mapped)  read    • Assembly    • Determining  genes/transcripts    

Posi5on  of  mapped  read    • Expression  levels  • Differen5al  expression    

5  

Page 6: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Benefits  and  opportuni5es  of  RNA-­‐seq  

•  All  transcripts  are  sequenced  not  just  ones  for  which  probes  are  designed  (cf  microarrays)  

•  Annota5on  of  new  exons,  transcribed  regions,  genes  or  non-­‐coding  RNAs  

•  Whole  transcriptome  sequencing  –  The  ability  to  look  at  alterna5ve  splicing  –  Allele  specific  expression  –  RNA  edi5ng  

Page 7: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

This  talk  

•  Analysis  of  RNA-­‐seq  data  for  the  purpose  of  determining  differen5al  expression  

•  How  much  are  expression  levels  changing  between  samples?  

Page 8: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

RNA-­‐seq  

Pepke  et  al,  Nature  Methods,  2009  

Page 9: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Raw  data  

•  Short  sequence  reads    •  Quality  scores  

@SEQ_ID!GATTTGGGGTTCAAAGCAGTATCGATCAAATAGTAAATCCATTTGTTCAACTCACAGTTT!+!!''*((((***+))%%%++)(%%%%).1***-+*''))**55CCF>>>>>>CCCCCCC65  

Page 10: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

RNA-­‐seq  analysis  steps  Raw  sequence  reads  

Map  onto  genome  

Summarize  reads  to  transcripts  

Sta5s5cal  tes5ng:  Determine  differen5ally  expressed  genes  

Systems  biology  

De  novo  assembly  Annota5on  based   Genome  guided  assembly  

Which  transcriptome?  

Page 11: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

RNA-­‐seq  analysis  steps  Raw  sequence  reads  

Map  onto  genome  

Summarize  reads  to  transcripts  

Sta5s5cal  tes5ng:  Determine  differen5ally  expressed  genes  

Systems  biology  

De  novo  assembly  Annota5on  based   Genome  guided  assembly  

Which  transcriptome?  

Page 12: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Map  reads  to  the  genome  

•  Aligning  millions  of  short  sequencing  onto  the  3  billion  bases  of  the  human  genome  

•  Accuracy  vs  speed  •  Many  aligners  available  (BWA,  Bow5e,  STAR,  Novoalign,…)  

Page 13: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

CDS   CDS   CDS   CDS  

CDS   CDS   CDS   CDS  

Gene  

transcript  

Sequencing  transcripts  not  the  genome  

Page 14: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Splice  site  mapping  •  Build  a  junc5on  library  from  all  combina5ons  of  known  exon  boundaries  

•  Determine  where  splice  junc5ons  occur  using  the  data  itself  -­‐  unbiased  by  annota5on.  

•  Several  so_ware  packages  to  do  this  such  as  TopHat,  SplitSeek,  SpliceMap…  

Gerber  et  al,  Nat  Methods,  2011  

Page 15: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Which  transcriptome  to  use?  

Page 16: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Raw  sequence  reads  

Map  onto  genome  

Summarize  reads  to  transcripts  

Sta5s5cal  tes5ng:  Determine  differen5ally  expressed  genes  

Systems  biology  

RNA-­‐seq  analysis  steps  

De  novo  assembly  Annota5on  based   Genome  guided  assembly  Which  transcriptome?  

Page 17: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Op5on  1  

•  Use  annota5on  (known  genes)  – Works  well  for  well  studied  organisms  (human,  mouse,  arabidopsis,  drosophila,…)    

– only  as  good  as  your  annota5on  – No  novel  transcripts  are  analysed  

Page 18: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Op5on  2:  Genome  guided  transcript  assembly  

•  Uses  the  loca5on  and  density  of  reads  along  the  genome  to  assemble  transcripts  

•  E.g.  Cufflinks  

•  Can’t  assemble  across  breaks  in  the  genome  – Cancer,  poor  genomes  

Page 19: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Op5on  3:  De  novo  transcriptome  assembly  

•  Assemble  transcripts  from  the  data  without  using  a  reference  genome  

•  “Harder”  than  genome  assembly  –  Orders  of  magnitude  varia5on  in  coverage  –  Con5gs  are  short  –  Alterna5ve  isoforms/transcripts  have  overlapping  sequences  –  *Very*  computa5onally  intensive  

•  So_ware  includes    –  Oases  (velvet)  –  TransAbyss  –  Trinity    –  …  

Page 20: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Example:  Annota5ng  the  chicken  W  chromosome  

   Z        Z      Z      W  

Male   Female  

Two  hypotheses  for  mechanisms  of  avian  sex  determina5on:  1.  Dominant  ovary  determining  gene  on  W  (cf  mammals)  2.  Dosage  of  Z-­‐linked  genes    

Page 21: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

There  is  an  annotated  chicken  genome  

•  Chicken  W  chromosome  is  poorly  assembled  •  Are  genes  on  other  chromosomes  really  on  the  W,  in  par5cular  the  random  chromosome?  

Chromosome Assembled Size (Mb)

Size inc. random (Mb)

Estimated Size (Mb)

Estimated Genes

(Ensembl) Z 69 70 80 796

W 0.24 0.89 18-54 46

Un_random 56 - - 1287

Page 22: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

                                                 Experimental  design      

   

PCR  Sexing  

+12hour  Blastoderms  

Hand  plate  for  PCR  Sexing  

RNA  

Stage  26  paired  gonads  (day  4.5)  

12  Female      

RNA  

16  Female  gonads  

Pooled  Samples  

12  Female  

16  Female  gonads  

12  Male  

12  Male  

16  Male  gonads  

16  Male  gonads  

RNA-­‐seq  • Illumina  HiSeq2000  • Paired-­‐end  100bp  • 4  lanes  • >80million  reads/sample  

 

Page 23: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Defining  the  transcriptome  

•  Annota5on  ~20,000  genes  •  Genome  guided  assembly  (Cufflinks)  ~45,000  genes  

•  De  novo  transcriptome  assembly  ~2.5  million  transcripts  (Abyss  with  filtering)!  

A  combined  approach  •  Assemble  cufflink  genes  using  transcripts  from  our  de  novo  assembly  

Page 24: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Annota5on  of  the  chicken  W  combined  all  three  approaches  

W/W_random ChromsomeUn_random ChromosomeAutosomes

Blastoderm CoverageGonads Coverage

Abyss TranscriptsCufflinks TranscriptsEnsembl Transcripts

1000 1500 2000 2500 3000 3500

base position

039

1

Coverage

Genome

EnsemblCufflinks

AbyssRASA1−W

1400 1600 1800 2000 2200

base position

012

6

Coverage

Genome

EnsemblCufflinks

AbyssST8SIA3−W

2000 2200 2400 2600 2800

base position

069

9

Coverage

Genome

EnsemblCufflinks

AbyssGOLPH3−W

0 500 1000 1500 2000 2500

base position

019

4

Coverage

Genome

EnsemblCufflinks

AbyssZSWIM6−W

0 200 400 600 800 1000 1200 1400

base position

082

Coverage

Genome

EnsemblCufflinks

AbyssNEDD4−like−W

Full  list  of  W  genes/transcripts  for  differen5al  expression  Ayers  et  al,  2013  

Page 25: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Summariza5on  

Take  your  “transcriptome”  and  add  up  the  reads  

Page 26: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Coding  Sequence   Exons   Introns   Splice  Junc5ons  

CDS   CDS   CDS   CDS  

§   Reads  in  exons  §   Exons  +  junc5ons  §   All  reads  start  to  end  of  transcript  §   De  novo  methods  

Page 27: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

CDS   CDS   CDS   CDS  

CDS   CDS   CDS   CDS  

CDS   CDS   CDS  

Even  when  all  transcripts  are  “known”  summariza5on  or  expression  quan5fica5on  is  difficult.  How  do  you  assign  reads  to  transcripts?  

Transcript  1  

Transcript  2  

CDS   CDS   CDS   CDS  Transcript  3  

Page 28: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Summariza5on  turns  mapped  reads  into  a  table  of  counts  

Tag  ID! A1! A2! B1! B2!ENSG00000124208! 478! 619! 4830! 7165!ENSG00000182463! 27! 20! 48! 55!ENSG00000125835! 132! 200! 560! 408!ENSG00000125834! 42! 60! 131! 99!ENSG00000197818! 21! 29! 52! 44!ENSG00000125831! 0! 0! 0! 0!ENSG00000215443! 4! 4! 9! 7!ENSG00000222008! 30! 23! 0! 0!ENSG00000101444! 46! 63! 54! 53!ENSG00000101333! 2256! 2793! 2702! 2976!

…" …  tens  of  thousands  more  tags  …"

**  very  high  dimensional  data  **  

Page 29: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

RNA-­‐seq  analysis  steps  Raw  sequence  reads  

Map  onto  genome  

Summarize  reads  to  transcripts  

Sta5s5cal  tes5ng:  Determine  differen5ally  expressed  genes  

Systems  biology  

De  novo  assembly  Annota5on  based   Genome  guided  assembly  

Which  transcriptome?  

Page 30: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Assessing  differen5al  expression  (DE)  

•  Which  genes  are  changing  in  their  abundance  between  samples?  

•  Sta5s5cal  tests  for  DE  (edgeR)  

Page 31: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Normaliza5on  

Accoun5ng  for/removing  technical  sources  of  varia5on  

Page 32: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Library  size  =  sum  of  each  column  

Tag  ID! A1! A2! B1! B2!ENSG00000124208! 478! 619! 4830! 7165!ENSG00000182463! 27! 20! 48! 55!ENSG00000125835! 132! 200! 560! 408!ENSG00000125834! 42! 60! 131! 99!ENSG00000197818! 21! 29! 52! 44!ENSG00000125831! 0! 0! 0! 0!ENSG00000215443! 4! 4! 9! 7!ENSG00000222008! 30! 23! 0! 0!ENSG00000101444! 46! 63! 54! 53!ENSG00000101333! 2256! 2793! 2702! 2976!

…" …  tens  of  thousands  more  tags  …"

Page 33: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Simple  thought  experiment  

•  Two  samples  A  and  B,  sequenced  to  the  same  depth  (same  library  size)  

•  Every  gene  that  is  expressed  in  A,  is  expressed  in  B  at  the  same  level  

•  Say  there  are  a  small  number  of  genes  that  are  expressed  uniquely  to  sample  B,  but  they  are  quite  highly  expressed  (lots  of  reads)  

•  Many  genes  within  the  common  set  will  appear  differen5ally  expressed  (B  <  A)  

Page 34: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Another  way  to  view  it  

•  Hypothe5cal  example:  Sequence  6  libraries  to  the  same  depth,  with  varying  levels  of  unique-­‐to-­‐sample  expression  

•  Differences  in  observed  counts  among  the  common  genes  

Red=low,  goldenyellow=high  

Page 35: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Marioni  et  al.  2008  RNA-­‐seq  data  

Distribu5on  of  log-­‐ra5os  of  counts    0  counts  are  omived    Red  line  =  trimmed  mean  

trimmed  mean  

•  5  lanes  of  liver  RNA,  5  lanes  of  kidney  RNA  •  Compare  two  single  kidney  libraries  (technical  replicates),  a_er  adjus5ng  for  library  size  

Page 36: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

.

(a)

log2(Kidney1 NK1) − log2(Kidney2 NK2)D

ensi

ty

-6 -4 -2 0 2 4 6

0.0

0.4

0.8

log2(Liver NL) - log2(Kidney NK)

Den

sity

-6 -4 -2 0 2 4 6

0.0

0.2

0.4(b)

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●●

●●

●●

●●

●●

●●

●●

●●

●●

●● ●

●●

●●

●●

● ●●

● ●

●●

●●

●●

●●

●●

●●●

● ●●

●●

●●

●●

● ●●

● ●●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

●●

● ●

●●

●●

●●

●●

●●

●●●

●●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

● ●

●●

● ●

●●

● ●

●●

●●

●●

● ●●

●●

●●

●●

● ●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●●

● ●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

● ●●

●●

●●

● ●

●●

●●●

●●

● ●

● ●

●●

●●

●●●

●●

●●

●●

●●

●●

●●

●● ●

● ●

●●

●●

● ●

● ●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

● ●

● ●●

●●

●● ●

●●

●●●

● ●

●●

●●●

●●

●●

●●

●●

●●

●●

●●

●●

●●●

●●

●●

●●

● ●● ●

●●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

● ●

●●

●●

● ●

●●

●●

●●

●● ●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●● ●●●

●●

● ●

●●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

●●

●●

● ●

● ●

●●● ●

●●

●●

●●

● ●

●●

● ●

●●

●●

● ●

●●

● ●

●● ●

● ●

●●

●●

●●

● ●●

●●

●●

●●

●●

●●

●●

● ●

● ●

● ●

●●

●● ●

●●

● ●

●●

● ●

●●

●●

●●

●●

●●

●●

● ● ●

● ●

● ●●

● ● ●

●●●

●● ●

●●

●●

● ●

●●

●●

●●

●●

● ●

●●

●●

● ●

●●

●●

● ●

●●

●●

●●

●●

● ●●

●● ●

●●

●●

● ●●

●●

●●

●●

● ●

●●

●●

●●

●●

● ●●

● ●

●●

●●

●●

●●

● ●

● ●

●●

●●

●●

●●●●

● ●

●●

●●

●● ●●

● ●

●●

●●

●●●

●●●

● ●

●●

● ●

●●

●●

● ●●

●●●

●●

●●

●●

●● ●●

● ●●

●●

●●

●●

●●

●●

●● ●

●●

●●●

●●

●●

●●

●●

● ●

●●

●●●

●● ●●● ● ●

●●●

●●

●● ●

● ●

● ●

●●●

● ●

● ●

●●

●●

●●

●●

●●

●●

●● ●

● ●

●●

●●

●●

● ●

● ●

●●

●●

● ●

● ●

●●

●● ● ●

●●

●●

●●

●●

●●

●●

●●

● ●● ●

● ●●

● ●●

●● ●

● ●

●●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

● ●

● ●

●●

●●

● ●

●●

● ●

●●

●●

● ●●

●●

●●

● ●●

● ●

●●

●●

●●

●●

● ●

●●

● ●

●●

●●●

●●●

●●

●●

●●

●●

●●

●●

●●●

●● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

●●

●●

●●

●●

●●

● ●●

●●

●●●

●●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●● ●

●●

●●

● ●

●●

●●

●●

● ●

●●

● ●

●●

● ●

●●

●●

●●

● ●

●●

● ●

● ●

●●

●●

● ●

●●

● ●

●●

●● ●

●●●●

●●●

●●

●●

●●

● ●

●●

●●

● ●●

● ●

●● ●●

●●

●●

●●

● ●

●●

●●

● ●

● ● ●

●●

●●

● ●●

●●

●●● ●

●●

●●

● ●

●●

●●

●●

●●

●●

●● ●

●●

●●●

●●●●

●●

● ●

●●

●●● ●

●●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●●

● ●

●● ●

● ●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●● ●

●●

●●

●●

●●

●●

●●

●●●

●●

●●

●●●

● ●

●●

●● ●

●●

●●

● ●

●●●

● ●●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

●●

● ●●

●●

●●

●●

●●

●●●

●● ●

● ●

● ●●

●●

●●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●●

● ●

●●

● ●

●●

●●

● ●

●●

● ●

●●

●●

●● ●●

●●

●●

●●

●●

●●

● ● ●●

●●

●●

●●

● ● ●●

●●

● ●

●●

●●

● ●

●●

● ●

●●

●●

● ●

●●

●●

● ●●

●●

●● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

●●

● ●

●●●●

●●

●● ●

●●

●●

●● ●●

●●

●●

● ●●

● ●

●●

●●

●●

●●

●●

● ●●

●●

● ●●

● ●●

●●

●●

●●

● ●

●●

●●

●●

●● ●●

●●

●●

●● ●

●● ●

● ●

●●

●●

●●

● ●

●● ●

●●

● ●●

● ●

●●

●●

● ● ●

●●

●●

●●

●●●

●●

●●

●●

●●

● ●

●●

●●

● ● ●

● ●

●●

●●

●●

●●

●●

● ●

● ●●

●●

● ●

●●● ●● ●

●●

●●

●●●

●●●

●●

●●

●●

●●●

●●

● ●

● ●

●●

●●

●●

●●●

● ● ●

●●

●●

●●

●● ●

●●●

● ●

●●●

●●

● ●●

●●

●●

● ●●

●●●●●

●●●

●●

●●

●●●

●●

●●

●●

● ●

●●

● ●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●● ●

●●

●●

● ●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●●

●●

● ●●

● ●

● ●

●●●

● ●●

● ●

●●

● ●

●●

●●

●●

●●

● ●

● ●

● ●

●●

●●

●●

● ●

● ●

●●

●●

●● ●

●●

●●

●●

● ●● ●

● ●●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

●●

●●●

●●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●● ●

●●

●●

●●

●●

●●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●●

● ●

● ● ●

●●

●●

●● ●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

● ●

●●●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●

● ●●

● ●

● ●●

●●

●●

● ●●

●●

● ●

● ●

●●

●●

●●

●●

●●

●●

●●

●●●

●●

●●

●●

●●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●●

●●

● ●

●●

●●

●●

●●

●●

●●

● ●●

●● ●●

●●

●●

●●

●●

●●

●●

● ●●

●●● ●

● ●

●●

●●

●●

● ●

●●

●●● ●

●●

● ●

●●

●●

●●

●● ●●●

●●

●●

●●

● ●

●●

●● ●

● ●●

●●

●●

●● ●

●●●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●●

● ●

●●

●● ●

● ●

●●

●●

●●

●●

● ●

● ●

●●

●●

●●

●●

●●

● ●●

●●

●● ●

●●

●●

●●

●●

● ●

●● ●●

●●

● ●

● ●

● ●

●●●

●●●

● ●

●●

● ●

●●

● ●

●●

●●

●● ●

● ●

●●

●●● ● ●

●●

●●

● ●

●●

● ●

●●

●●

●●

●●●

●●

●●

●●● ● ●●●

●●

● ●●

●●●

● ● ●●

●●

●●

●●●●

●●

●● ●

●●

●●

● ●●

●●

●●●●

● ●●●

●●

●●

●●

●● ●

● ●●

●●

●●

●●

●●

●●●

● ● ●

●●

●●

● ●

●●

● ●

●●

●●●

●●

●●

●●

●●

●●

●●●

●●

● ●

●●

● ●

●●●

●●

● ●

●●

●●

● ●

●●

●●

●●

● ●

●●

● ●

●●

● ●

●●

●● ●

●●

●●

●●

●●

● ●

●●

●● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●●

●●

● ●

● ●

● ●

● ●●

●●

●●

● ●

●●

●●

●●

●●

●●●

●●

●●

●●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●● ●

●●

●●

●●

●●

●●

● ●

●●

●●

●● ●

● ●

●●

●●

●●

●●

●●

●● ●●● ●

● ●

● ●

●●●

● ●

●●

●●

●●

● ●●●

●●●●

●● ●

● ● ●

●● ●

●●

●● ●

●●

●●

●●

●●

●●●●

● ●

● ●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●●

●●

●●

●●

●●

●●

●●

●● ●

●●●

● ● ●●

●●

●●

●●

●●

●● ●

●●

●●

●●

●●

●● ●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●●

●●

●●

●●

●●

● ●

●●

● ●

●●

●●

●●

●●

●●

●● ●

●●●

●●

●●

●●

● ●

●●

●●

●●

●● ●

●● ●

●●

●●

● ●●

●●

●●

●●

● ●●

●●

●●

●●

●●

●●●

●●

● ●

● ●

●●

●●

●●●

●●

● ●

●●●

●●

●●

●●

●●

● ●●

● ●

●●

●●

●●

●●

●●

● ●

●●

●●

● ●

●●

●●

● ●

●●

● ●

●●

● ●

●●

●●

●●

●●

●●

● ●

● ●

●●

● ●

● ●

●● ●

●●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

● ●

● ●

●●●●

●●

●●

●●

●●

●●

●●●

● ●

●●

●●

●●

●●

● ●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●

●● ●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

●●

●●

● ●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

● ●

●●

●●●

●●

●●●

● ●

●●

●●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

● ●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●●

●●

●●

● ●

●●

●●

●●

●● ●

●●

●●

● ●

● ●

●●

●●

● ●

● ●

●●

●●

●●

●●

● ●

● ●

●●●

●●

●●

●●

●●

●● ●

●●

● ●

●●

●●

● ●

●●

●●

● ●

● ●

●●●

●●

●●

●●

●●

● ●●

● ●

●●

●●●

●●

● ●

● ●

●●

● ●

●●

●●

●●

●● ●

●●

●●

● ●

●●

●●

●●

●●● ●

●●

●●

●●

●●

●●

●● ●

●●

●●

●●

●●

● ●

●●

●●

● ●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

● ●

●●●

●●

●●

● ●

● ●

●●

●●

● ●

●●

●●●●

●●

●●

●●

●● ●

● ●

●●

●● ●

●●

● ●

●●

●●●

●●

●●

● ●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

● ●

●●

●●

●●

● ●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●● ●

●●

●●

●●●

●● ●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

● ●

● ●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ● ●

●● ●

●●

●●

● ●● ●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

● ●

● ●●

●●

●●

●●

●●

● ●

●●●

●●

● ●

●●

●●

●●

● ●

● ●

●●

●●

●●

● ●

●●

● ●

●●●

●● ●

● ●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

● ●●

● ●

● ●

●●

●●

●●

●●

●●

●●

●●

●●●

●●

● ●

●●

●●●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

●●

●●

●●

●●

●● ●

●●●● ●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

●●

●●

●● ●

● ●

● ●

●● ●●

●●

●●●

●●

●●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●●

●●

●●

●●

●●●

● ●● ●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●

● ●

●●

●●●

●●

●●

●●

● ●●

● ●

● ●●●

●●

●●

●●

●●

●●

●●●

●●

● ●

●●

● ●

●●

●●

●●

●●

● ●

●●●

●●

●●

●●

●●

●● ●

●●

●●

● ●●●

●●

● ●

●●

●●

● ●

● ●●

●●

●●

●●

●●

●●

●●●

●● ●● ●

● ●●

●●

●● ●● ●●

● ●

●●

● ●

●●

●●

●●

●●

●●

● ●

● ●

●●●

●● ●

●●

● ●

●●

● ●

● ●●

●●

●●

●●

●●

●● ●

●●

●●

● ●

●●

●●

● ●

●●

●● ●

●●

● ● ●

●●

●●

●●

● ●

●●

●●

● ●

●●

●●

●●

●●

● ●●

●●

● ●

● ●●

●●

●●

● ●

●●

●●

●●

● ●●

●●

● ●

● ●

●●

●●

●●●

●●

●● ●

●●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

● ●

● ●

●●

● ●

●●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●● ●

● ●

● ●

●●

●●

●●

●●

●●●

●●●

● ●●●● ●●

●●

●●

●●

●●● ● ●

● ●

●●●

●●

● ●

●● ●

●●

●●

●●●

●●

●●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●●

●●

●●

●●

●●●

●●

●●

●●

● ● ●

●●

●●

● ●

●● ●

● ●

● ●

●●

● ●

●●

●●

● ●●

●●

●●

●●

●●

●●

● ●

● ●

●●●

●●●●●

●●

●●

●●

●●

● ●

● ●

●●

●●

●●

● ●

● ●

●● ●●

●●

●●

● ●

● ●

●●

●●

● ●

●●●●

●●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

●●

●●

●●

● ●●

●●

●●

●●

● ●

● ●

●●

● ●

● ●

●●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

● ●

●●

●●●

●●

●●●

●●

● ●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

●●

●●

●●

● ●●●●

●●●

●●

●● ●

● ●

● ● ●

●●

●●

●● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●●

●●

● ●●

●●

● ●

●●

●●

●●

● ●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

● ●●

●●

●●●

●●

●●

●●

● ●

● ●

●●

●●

●●

●●

●●●●

●●●

● ●●

● ●

●●

●● ●●

●●

● ●●

●● ●

●●●

●●

●●●

●●●

● ●

● ●

●●

● ●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

●●

●●

● ●

● ●●

●●

● ●●

●●

●●

●●

●●

●●

●●

●●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●●

● ●

●●

● ●

●●

●● ●

●●

●●

●●

●●

●●

●●

● ●

● ●

●●

●●

●●

●●

●●

●●

●●

●● ●

●●

●●

●●

●●● ●

●●

●●

●●

● ●

● ●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●●

●●

●●

●●

●●

●●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●● ●

●●●●●

●●

●●

●●

●●●

●●

●●●

● ●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

●●

● ●●

●●

● ●●

●●

●●

●●

● ●

●●

● ●

● ●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●●

● ●

●●

●●

●●

●●

●●

●●

● ●

●●

●● ●

●●

●●

●●

● ●

●●

● ●

●●

●●

●●

● ●●

●●

● ●

● ●

●●

●●

●●

●●

●●

● ●

●● ●

●●

● ●

●●

●●

● ●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●●

●●

●●

●●

● ●

● ●

●●

● ●

● ●

● ●

●● ● ●

●●

●●●

●●

● ●●

●●

●●

●●

● ●●●

●●

● ●●

● ●

●●

● ●

● ●

●●

●●

●●

●●

● ●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●●

●●

●●

●●●

●●

●● ● ●

● ●●

● ●

●●

●●

●●

● ●

● ●

● ●●

●●

●●

●●

●●

●●

●●

● ●

●●●

●●

●● ●

● ●

●●

●●

●●

●●

● ●

●●

● ●

●●

●●

●●

●●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●●

●●

●●

●●

●●●

●●

●●

●●●

●● ●

●●

●●

●●

●●

●●

●●

● ●● ●

●●

●●

●●

● ●●

●●

● ●

●●

●●

●●

● ●

●●

●●●

●●

●●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●● ●

●●

●●

●●●● ●●

●●

● ●

● ●● ● ●●

● ●

●●

●●

●●

●●●●

●●

●●

●● ●

● ●●

●●

●●

●●

●●●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●●

● ●

●●

●●

●● ●●

● ●

●● ●

●●●

● ●

● ●

●●

●●

●●

●● ●

● ● ●

●●

● ●

●●

●●

●●

● ● ●●

●●

●●●

●●

● ●

●●

●●

● ● ●

●●

●●

●●

●●

●●●

●●

●●

●●

● ●●

●●

●● ●

●●

●●

●●

●●

● ●● ●

●●

●●

●●

●●

● ●●

●●

●●

●●

●●

●●

● ●

●●●

●●

● ●● ● ●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

● ●

●●●

●●●●

●● ●

●● ●

●●

●●

●●

●● ●

●●

●●

●●

● ●

●●

●●● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●●

●●●

●●

●●●

●●

● ●

●●

●●

●●

● ●

●●

●● ●

●●

● ●●

●●

● ●

●●

●●

●●

● ●

●●

● ●

●●

●●●

●●

●●

● ●

●●

● ●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●●

●●

●●

●●

●● ●

● ●

●●

●●

●●●

●●

●●●

●●

● ●

● ●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

●●

●●

●● ●

●●

●●

● ● ●

●●

-20 -15 -10

-50

5

A = log2( Liver NL Kidney NK)

M=

log 2

(Liv

erN

L)-

log 2

(Kid

ney

NK)

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●●

●●

●●

●●

●●

●●

●●●

●● ●

●●

●●

● ●

●●●

●●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

●●

●●●

●●

● ●●

● ●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

Housekeeping genesUnique to a sample

(c)

Same  sample.  Technical  replicates  

Different  samples.  Liver  vs  Kidney  

Page 37: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Another  way  to  view  the  data  

.

(a)

log2(Kidney1 NK1) − log2(Kidney2 NK2)

Den

sity

-6 -4 -2 0 2 4 6

0.0

0.4

0.8

log2(Liver NL) - log2(Kidney NK)

Den

sity

-6 -4 -2 0 2 4 6

0.0

0.2

0.4(b)

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●●

●●

●●

●●

●●

●●

●●

●●

●●

●● ●

●●

●●

●●

● ●●

● ●

●●

●●

●●

●●

●●

●●●

● ●●

●●

●●

●●

● ●●

● ●●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

●●

● ●

●●

●●

●●

●●

●●

●●●

●●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

● ●

●●

● ●

●●

● ●

●●

●●

●●

● ●●

●●

●●

●●

● ●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●●

● ●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

● ●●

●●

●●

● ●

●●

●●●

●●

● ●

● ●

●●

●●

●●●

●●

●●

●●

●●

●●

●●

●● ●

● ●

●●

●●

● ●

● ●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

● ●

● ●●

●●

●● ●

●●

●●●

● ●

●●

●●●

●●

●●

●●

●●

●●

●●

●●

●●

●●●

●●

●●

●●

● ●● ●

●●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

● ●

●●

●●

● ●

●●

●●

●●

●● ●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●● ●●●

●●

● ●

●●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

●●

●●

● ●

● ●

●●● ●

●●

●●

●●

● ●

●●

● ●

●●

●●

● ●

●●

● ●

●● ●

● ●

●●

●●

●●

● ●●

●●

●●

●●

●●

●●

●●

● ●

● ●

● ●

●●

●● ●

●●

● ●

●●

● ●

●●

●●

●●

●●

●●

●●

● ● ●

● ●

● ●●

● ● ●

●●●

●● ●

●●

●●

● ●

●●

●●

●●

●●

● ●

●●

●●

● ●

●●

●●

● ●

●●

●●

●●

●●

● ●●

●● ●

●●

●●

● ●●

●●

●●

●●

● ●

●●

●●

●●

●●

● ●●

● ●

●●

●●

●●

●●

● ●

● ●

●●

●●

●●

●●●●

● ●

●●

●●

●● ●●

● ●

●●

●●

●●●

●●●

● ●

●●

● ●

●●

●●

● ●●

●●●

●●

●●

●●

●● ●●

● ●●

●●

●●

●●

●●

●●

●● ●

●●

●●●

●●

●●

●●

●●

● ●

●●

●●●

●● ●●● ● ●

●●●

●●

●● ●

● ●

● ●

●●●

● ●

● ●

●●

●●

●●

●●

●●

●●

●● ●

● ●

●●

●●

●●

● ●

● ●

●●

●●

● ●

● ●

●●

●● ● ●

●●

●●

●●

●●

●●

●●

●●

● ●● ●

● ●●

● ●●

●● ●

● ●

●●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

● ●

● ●

●●

●●

● ●

●●

● ●

●●

●●

● ●●

●●

●●

● ●●

● ●

●●

●●

●●

●●

● ●

●●

● ●

●●

●●●

●●●

●●

●●

●●

●●

●●

●●

●●●

●● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

●●

●●

●●

●●

●●

● ●●

●●

●●●

●●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●● ●

●●

●●

● ●

●●

●●

●●

● ●

●●

● ●

●●

● ●

●●

●●

●●

● ●

●●

● ●

● ●

●●

●●

● ●

●●

● ●

●●

●● ●

●●●●

●●●

●●

●●

●●

● ●

●●

●●

● ●●

● ●

●● ●●

●●

●●

●●

● ●

●●

●●

● ●

● ● ●

●●

●●

● ●●

●●

●●● ●

●●

●●

● ●

●●

●●

●●

●●

●●

●● ●

●●

●●●

●●●●

●●

● ●

●●

●●● ●

●●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●●

● ●

●● ●

● ●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●● ●

●●

●●

●●

●●

●●

●●

●●●

●●

●●

●●●

● ●

●●

●● ●

●●

●●

● ●

●●●

● ●●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

●●

● ●●

●●

●●

●●

●●

●●●

●● ●

● ●

● ●●

●●

●●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●●

● ●

●●

● ●

●●

●●

● ●

●●

● ●

●●

●●

●● ●●

●●

●●

●●

●●

●●

● ● ●●

●●

●●

●●

● ● ●●

●●

● ●

●●

●●

● ●

●●

● ●

●●

●●

● ●

●●

●●

● ●●

●●

●● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

●●

● ●

●●●●

●●

●● ●

●●

●●

●● ●●

●●

●●

● ●●

● ●

●●

●●

●●

●●

●●

● ●●

●●

● ●●

● ●●

●●

●●

●●

● ●

●●

●●

●●

●● ●●

●●

●●

●● ●

●● ●

● ●

●●

●●

●●

● ●

●● ●

●●

● ●●

● ●

●●

●●

● ● ●

●●

●●

●●

●●●

●●

●●

●●

●●

● ●

●●

●●

● ● ●

● ●

●●

●●

●●

●●

●●

● ●

● ●●

●●

● ●

●●● ●● ●

●●

●●

●●●

●●●

●●

●●

●●

●●●

●●

● ●

● ●

●●

●●

●●

●●●

● ● ●

●●

●●

●●

●● ●

●●●

● ●

●●●

●●

● ●●

●●

●●

● ●●

●●●●●

●●●

●●

●●

●●●

●●

●●

●●

● ●

●●

● ●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●● ●

●●

●●

● ●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●●

●●

● ●●

● ●

● ●

●●●

● ●●

● ●

●●

● ●

●●

●●

●●

●●

● ●

● ●

● ●

●●

●●

●●

● ●

● ●

●●

●●

●● ●

●●

●●

●●

● ●● ●

● ●●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

●●

●●●

●●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●● ●

●●

●●

●●

●●

●●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●●

● ●

● ● ●

●●

●●

●● ●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

● ●

●●●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●

● ●●

● ●

● ●●

●●

●●

● ●●

●●

● ●

● ●

●●

●●

●●

●●

●●

●●

●●

●●●

●●

●●

●●

●●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●●

●●

● ●

●●

●●

●●

●●

●●

●●

● ●●

●● ●●

●●

●●

●●

●●

●●

●●

● ●●

●●● ●

● ●

●●

●●

●●

● ●

●●

●●● ●

●●

● ●

●●

●●

●●

●● ●●●

●●

●●

●●

● ●

●●

●● ●

● ●●

●●

●●

●● ●

●●●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●●

● ●

●●

●● ●

● ●

●●

●●

●●

●●

● ●

● ●

●●

●●

●●

●●

●●

● ●●

●●

●● ●

●●

●●

●●

●●

● ●

●● ●●

●●

● ●

● ●

● ●

●●●

●●●

● ●

●●

● ●

●●

● ●

●●

●●

●● ●

● ●

●●

●●● ● ●

●●

●●

● ●

●●

● ●

●●

●●

●●

●●●

●●

●●

●●● ● ●●●

●●

● ●●

●●●

● ● ●●

●●

●●

●●●●

●●

●● ●

●●

●●

● ●●

●●

●●●●

● ●●●

●●

●●

●●

●● ●

● ●●

●●

●●

●●

●●

●●●

● ● ●

●●

●●

● ●

●●

● ●

●●

●●●

●●

●●

●●

●●

●●

●●●

●●

● ●

●●

● ●

●●●

●●

● ●

●●

●●

● ●

●●

●●

●●

● ●

●●

● ●

●●

● ●

●●

●● ●

●●

●●

●●

●●

● ●

●●

●● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●●

●●

● ●

● ●

● ●

● ●●

●●

●●

● ●

●●

●●

●●

●●

●●●

●●

●●

●●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●● ●

●●

●●

●●

●●

●●

● ●

●●

●●

●● ●

● ●

●●

●●

●●

●●

●●

●● ●●● ●

● ●

● ●

●●●

● ●

●●

●●

●●

● ●●●

●●●●

●● ●

● ● ●

●● ●

●●

●● ●

●●

●●

●●

●●

●●●●

● ●

● ●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●●

●●

●●

●●

●●

●●

●●

●● ●

●●●

● ● ●●

●●

●●

●●

●●

●● ●

●●

●●

●●

●●

●● ●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●●

●●

●●

●●

●●

● ●

●●

● ●

●●

●●

●●

●●

●●

●● ●

●●●

●●

●●

●●

● ●

●●

●●

●●

●● ●

●● ●

●●

●●

● ●●

●●

●●

●●

● ●●

●●

●●

●●

●●

●●●

●●

● ●

● ●

●●

●●

●●●

●●

● ●

●●●

●●

●●

●●

●●

● ●●

● ●

●●

●●

●●

●●

●●

● ●

●●

●●

● ●

●●

●●

● ●

●●

● ●

●●

● ●

●●

●●

●●

●●

●●

● ●

● ●

●●

● ●

● ●

●● ●

●●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

● ●

● ●

●●●●

●●

●●

●●

●●

●●

●●●

● ●

●●

●●

●●

●●

● ●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●

●● ●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

●●

●●

● ●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

● ●

●●

●●●

●●

●●●

● ●

●●

●●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

● ●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●●

●●

●●

● ●

●●

●●

●●

●● ●

●●

●●

● ●

● ●

●●

●●

● ●

● ●

●●

●●

●●

●●

● ●

● ●

●●●

●●

●●

●●

●●

●● ●

●●

● ●

●●

●●

● ●

●●

●●

● ●

● ●

●●●

●●

●●

●●

●●

● ●●

● ●

●●

●●●

●●

● ●

● ●

●●

● ●

●●

●●

●●

●● ●

●●

●●

● ●

●●

●●

●●

●●● ●

●●

●●

●●

●●

●●

●● ●

●●

●●

●●

●●

● ●

●●

●●

● ●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

● ●

●●●

●●

●●

● ●

● ●

●●

●●

● ●

●●

●●●●

●●

●●

●●

●● ●

● ●

●●

●● ●

●●

● ●

●●

●●●

●●

●●

● ●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

● ●

●●

●●

●●

● ●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●● ●

●●

●●

●●●

●● ●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

● ●

● ●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ● ●

●● ●

●●

●●

● ●● ●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

● ●

● ●●

●●

●●

●●

●●

● ●

●●●

●●

● ●

●●

●●

●●

● ●

● ●

●●

●●

●●

● ●

●●

● ●

●●●

●● ●

● ●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

● ●●

● ●

● ●

●●

●●

●●

●●

●●

●●

●●

●●●

●●

● ●

●●

●●●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

●●

●●

●●

●●

●● ●

●●●● ●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

●●

●●

●● ●

● ●

● ●

●● ●●

●●

●●●

●●

●●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●●

●●

●●

●●

●●●

● ●● ●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●

● ●

●●

●●●

●●

●●

●●

● ●●

● ●

● ●●●

●●

●●

●●

●●

●●

●●●

●●

● ●

●●

● ●

●●

●●

●●

●●

● ●

●●●

●●

●●

●●

●●

●● ●

●●

●●

● ●●●

●●

● ●

●●

●●

● ●

● ●●

●●

●●

●●

●●

●●

●●●

●● ●● ●

● ●●

●●

●● ●● ●●

● ●

●●

● ●

●●

●●

●●

●●

●●

● ●

● ●

●●●

●● ●

●●

● ●

●●

● ●

● ●●

●●

●●

●●

●●

●● ●

●●

●●

● ●

●●

●●

● ●

●●

●● ●

●●

● ● ●

●●

●●

●●

● ●

●●

●●

● ●

●●

●●

●●

●●

● ●●

●●

● ●

● ●●

●●

●●

● ●

●●

●●

●●

● ●●

●●

● ●

● ●

●●

●●

●●●

●●

●● ●

●●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

● ●

● ●

●●

● ●

●●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●● ●

● ●

● ●

●●

●●

●●

●●

●●●

●●●

● ●●●● ●●

●●

●●

●●

●●● ● ●

● ●

●●●

●●

● ●

●● ●

●●

●●

●●●

●●

●●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●●

●●

●●

●●

●●●

●●

●●

●●

● ● ●

●●

●●

● ●

●● ●

● ●

● ●

●●

● ●

●●

●●

● ●●

●●

●●

●●

●●

●●

● ●

● ●

●●●

●●●●●

●●

●●

●●

●●

● ●

● ●

●●

●●

●●

● ●

● ●

●● ●●

●●

●●

● ●

● ●

●●

●●

● ●

●●●●

●●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

●●

●●

●●

● ●●

●●

●●

●●

● ●

● ●

●●

● ●

● ●

●●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

● ●

●●

●●●

●●

●●●

●●

● ●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

●●

●●

●●

● ●●●●

●●●

●●

●● ●

● ●

● ● ●

●●

●●

●● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●●

●●

● ●●

●●

● ●

●●

●●

●●

● ●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

● ●●

●●

●●●

●●

●●

●●

● ●

● ●

●●

●●

●●

●●

●●●●

●●●

● ●●

● ●

●●

●● ●●

●●

● ●●

●● ●

●●●

●●

●●●

●●●

● ●

● ●

●●

● ●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

●●

●●

● ●

● ●●

●●

● ●●

●●

●●

●●

●●

●●

●●

●●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●●

● ●

●●

● ●

●●

●● ●

●●

●●

●●

●●

●●

●●

● ●

● ●

●●

●●

●●

●●

●●

●●

●●

●● ●

●●

●●

●●

●●● ●

●●

●●

●●

● ●

● ●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●●

●●

●●

●●

●●

●●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●● ●

●●●●●

●●

●●

●●

●●●

●●

●●●

● ●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

●●

● ●●

●●

● ●●

●●

●●

●●

● ●

●●

● ●

● ●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●●

● ●

●●

●●

●●

●●

●●

●●

● ●

●●

●● ●

●●

●●

●●

● ●

●●

● ●

●●

●●

●●

● ●●

●●

● ●

● ●

●●

●●

●●

●●

●●

● ●

●● ●

●●

● ●

●●

●●

● ●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●●

●●

●●

●●

● ●

● ●

●●

● ●

● ●

● ●

●● ● ●

●●

●●●

●●

● ●●

●●

●●

●●

● ●●●

●●

● ●●

● ●

●●

● ●

● ●

●●

●●

●●

●●

● ●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●●

●●

●●

●●●

●●

●● ● ●

● ●●

● ●

●●

●●

●●

● ●

● ●

● ●●

●●

●●

●●

●●

●●

●●

● ●

●●●

●●

●● ●

● ●

●●

●●

●●

●●

● ●

●●

● ●

●●

●●

●●

●●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●●

●●

●●

●●

●●●

●●

●●

●●●

●● ●

●●

●●

●●

●●

●●

●●

● ●● ●

●●

●●

●●

● ●●

●●

● ●

●●

●●

●●

● ●

●●

●●●

●●

●●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●● ●

●●

●●

●●●● ●●

●●

● ●

● ●● ● ●●

● ●

●●

●●

●●

●●●●

●●

●●

●● ●

● ●●

●●

●●

●●

●●●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●●

● ●

●●

●●

●● ●●

● ●

●● ●

●●●

● ●

● ●

●●

●●

●●

●● ●

● ● ●

●●

● ●

●●

●●

●●

● ● ●●

●●

●●●

●●

● ●

●●

●●

● ● ●

●●

●●

●●

●●

●●●

●●

●●

●●

● ●●

●●

●● ●

●●

●●

●●

●●

● ●● ●

●●

●●

●●

●●

● ●●

●●

●●

●●

●●

●●

● ●

●●●

●●

● ●● ● ●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

● ●

●●●

●●●●

●● ●

●● ●

●●

●●

●●

●● ●

●●

●●

●●

● ●

●●

●●● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●●

●●●

●●

●●●

●●

● ●

●●

●●

●●

● ●

●●

●● ●

●●

● ●●

●●

● ●

●●

●●

●●

● ●

●●

● ●

●●

●●●

●●

●●

● ●

●●

● ●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●●

●●

●●

●●

●● ●

● ●

●●

●●

●●●

●●

●●●

●●

● ●

● ●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

● ●

●●

●●

●●

●● ●

●●

●●

● ● ●

●●

-20 -15 -10

-50

5

A = log2( Liver NL Kidney NK)

M=

log 2

(Liv

erN

L)-

log 2

(Kid

ney

NK)

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●●

●●

●●

●●

●●

●●

●●●

●● ●

●●

●●

● ●

●●●

●●

●●

●●

●●

● ●

●●

●●

●●

● ●

●●

●●

●●●

●●

● ●●

● ●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●●

●●

●●

●●

●●

● ●

●●

●●

●●

●●

●●

●●

●●

●●

●●

● ●

●●

●●

●●

Housekeeping genesUnique to a sample

(c)

•  Sequencing  “real  estate”  is  fixed.  

•  Underlying  RNA  composi5on  can  be  very  different  –  e.g.  several  liver-­‐specific  genes  

•  An  adjustment  at  the  analysis  stage  should  be  made  

 

Page 38: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

The  adjustment  to  data  analysis  is  a  scaling  constant  

•  Assump5on:  core  set  of  genes  that  do  not  change  in  expression.  

•  Pick  a  reference  sample,  compute  TMM  rela5ve  to  reference  •  TMM  (Trimmed  Mean  of  M  values)  M=log  ra5o    

•  Adjustment  to  sta5s5cal  analysis:  –  Use  addi5onal  offset  (GLM)  

Robinson  &  Oshlack,  Genome  Biology,  2010  

Page 39: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Sta5s5cal  tes5ng  for  DE  

•  For  EACH  GENE,  is  the  mean  expression  level  for  the  gene  under  one  condi5on  significantly  different  from  the  mean  expression  level  under  a  different  condi5on?  

Tag  ID! A1! A2! B1! B2!ENSG00000124208! 478! 619! 4830! 7165!ENSG00000182463! 27! 20! 48! 55!ENSG00000125835! 132! 200! 560! 408!ENSG00000125834! 42! 60! 131! 99!

…" …  tens  of  thousands  more  tags  …"

39  

Page 40: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

DE  analysis  happens  in  R  •  Several  packages  currently  available  for  DE  analysis  of  RNA-­‐seq  data  in  R:  –  baySeq,    –  DEGSeq,    –  DESeq    –  edgeR  –  NBPSeq    (+  more  methods  with  less  refined  code,  e.g.  TSPM)  

•  edgeR  developed  at  WEHI,  so  this  is  my  favourite  •  edgeR  is  only  package  so  far  with  full  GLM  capabili5es  

•  Great  support  on  Bioconductor  mailing  list!  

Page 41: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

41  

Counts  from  a  Single  RNA  Sample  Can  be  Modeled  with  Poisson  Distribu5on  

RNA  sample  

M  =  total  number  of  reads  ≈  20  million  (for  example)  λg    =  true  propor5on  of  gene  g  yg    =  number  of  reads  for  gene  g      

Read  1  Read  2  Read  3  Read  4  Read  5  Read  6  …  

Short  reads  

Map  reads  to  genome  

Davis  McCarthy  

Page 42: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

42  

Counts  from  a  Single  RNA  Sample  Can  be  Modeled  with  Poisson  Distribu5on  

RNA  sample  

M  =  total  number  of  reads  ≈  20  million  λg    =  true  propor5on  of  gene  g  yg    =  number  of  reads  for  gene  g      

Large  M,  small  λi  à  yg  approximately  Poisson,  μg  =  Mλg  

Read  1  Read  2  Read  3  Read  4  Read  5  Read  6  …  

Short  reads  

Map  reads  to  genome  

What  we’re  interested  in  

What  we  observe  

Davis  McCarthy  

Page 43: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

43  

A  Small  RNA-­‐Seq  Experiment  (Tech  Reps)  

Condi5on  A   Condi5on  B  

λg1   λg2   λg3   λg4  

yg1   yg2   yg3   yg4  

Genes  g  =  1,  …,  30k  

M1  

M2  M3   M4  

E(ygi)  =  Mi  λgi  Reads  Mi  ≈  20  million  

Page 44: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

True  Technical  Reps  Show  Poisson  Varia5on  for  Each  Gene  

Data:    Marioni  et  al.,  Genome  Res,  2008   44  

Davis  McCarthy  

Page 45: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

45  

A  Small  RNA-­‐Seq  Experiment  (Biological  Reps)  

Condi5on  A   Condi5on  B  

λg1   λg2   λg3   λg4  

yg1   yg2   yg3   yg4  

Genes  g  =  1,  …,  30k  

M1  

M2  M3   M4  

E(ygi)  =  Mi  λgi  Reads  Mi  ≈  20  million  

Davis  McCarthy  

Page 46: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Biological  Replicate  Data  shows  Quadra5c  Mean-­‐Variance  Rela5onship  

(development  cycle  of  slime  mould,  2  samples  at  hr00,  &  2  at  hr04)  

binned  variance,  sample  variance  

Data:  Parikh  et  al,  Genome  Biology,  2010  

46  Davis  McCarthy  

Page 47: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Need  to  account  for  biological  varia5on  

•  Technical  replicate  counts  for  a  gene  vary  according  to  a  Poisson  law,  i.e.  sequencing  varia5on  is  Poisson  

•  Biological  CV  (BCV)  is  the  coefficient  of  varia5on  with  which  the  (unknown)  true  abundance  of  the  gene  varies  between  RNA  samples.    

•  edgeR  models  a  quadra5c  mean-­‐variance  rela5onship  var(ygi)=  μgi  +  φgμgi

2  with  φg=Biological  CV2  

Robinson  et  al,  2010  

Page 48: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Output  of  analysis  

•  A  list  of  genes  with  p-­‐values  for  tes5ng  differen5al  expression  between  samples  

•  A  ranked  list  of  genes  

Page 49: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

RNA-­‐seq  analysis  steps  Raw  sequence  reads  

Map  onto  genome  

Summarize  reads  to  transcripts  

Sta5s5cal  tes5ng:  Determine  differen5ally  expressed  genes  

Systems  biology  

De  novo  assembly  Annota5on  based   Genome  guided  assembly  

Which  transcriptome?  

Page 50: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

RNA-­‐seq  analysis  steps  Raw  sequence  reads  

Map  onto  genome  

Summarize  reads  to  transcripts  

Sta5s5cal  tes5ng:  Determine  differen5ally  expressed  genes  

Systems  biology  

De  novo  assembly  Annota5on  based   Genome  guided  assembly  

Which  transcriptome?  

Learn  something!  

Page 51: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

The  future  

•  Effec5vely  dealing  with  no  genome  or  poor  quality  genomes  (our  new  method,  Corset)  

•  Integra5ng  RNA-­‐seq  data  with  other  genomics  data  

•  Analysis  methodology  is  cri5cal  and  s5ll  developing  for  specific  purposes  

•  Opportuni5es  to  use  this  data  in  new  and  imagina5ve  ways  –  requires  new  analysis  methodology  

Page 52: The RNA-‐SEQ Analysis Pipeline - Alicia Oshlack

Acknowledgements  

•  Mark  Robinson  (Uni  Zurich)  •  Mavhew  Young  (Cambridge)  •  Nadia  Davidson  (MCRI)  •  Mavhew  Wakefield  (WEHI)  •  Gordon  Smyth  (WEHI)  •  Davis  McCarthy  (Oxford)