1. Dr. Christoph W. Sensen und Dr. Jung Soh Trieste Course 2017
RNA-Seq
Dr. Christoph W. Sensen und Dr. Jung Soh
Trieste Course 2017

2. What is RNA-Seq?
An experimental protocol that uses next-generation sequencing technologies to sequence the messenger RNA molecules within a biological sample in an effort to determine the primary sequence and relative abundance of each mRNA
Martin JA, Wang Z (2011) Next-generation transcriptome assembly. Nat Rev Genet. 12(10):
Also known as “Whole Transcriptome Shotgun Sequencing” (WTSS)

3. Sequencing strategy Metabolite profiling Plant material
combination of ½-plate of 454 and 1 lane of 108PE Illumina sequencing
excellent depth and coverage
high-quality assemblies
submission of total RNA samples improves quality control
takes better advantage of sequencing facilities
similar overall cost
76SE Illumina sequencing on selected species for comparative transcriptomics
Plant material
Biochemistry PIs
Total RNA extraction
Bioanalyzer
(RNA quality)
mRNA isolation
cDNA libraries
Genome Québec Innovation Centre
454
(1/2-plate)
Illumina
1 lane 108PE
Reference transcriptomes (75)
repeat sequencing in rare cases of low-quality initial output
Bioinformatics
Innovation
Centre
Bioinformatics

4. RNA-Seq workflow intron
Wang Z, Gerstein M, Snyder M (2009) RNA-Seq: a revolutionary tool for transcriptomics. Nat Rev Genet. 10(1):57-63.

5. RNA-Seq vs. microarray Characteristics RNA-Seq Microarray
Which transcripts?
All in a sample
Only those for which probes are designed
Transcript sequence generation
Yes No
Low-abundance transcript detection
Limited
Abundance info source
Count (of the reads aligned to gene)
Fluorescence level (of the probe spot for gene)
Resolution
Base
Probe sequence
Background noise
Low
High
Additional info
Alternative splicing, transcriptome-level variation

6. RNA-Seq data analysis Map reads Bin reads to features Normalize counts
Lots of short reads
Reference genome
Map reads
Table of mapped loci per read
Feature annotation
(exons, genes, transcripts)
Bin reads to features
Table of counts per feature
Usually combined in a tool
Normalize counts
Table of normalized quantification values per feature
Detect differentially expressed (DE) features
DE features

7. Mapping reads Need a reference genome Issues Huge amounts of data
Reads spanning across exon junction
Alternative splicing
Reads mapping to multiple locations in the genome
Huge amounts of data
Most common mapping results format
SAM: sequence alignment/map
BAM: binary format of SAM
Many tools
Bowtie, SOAP, BWA, SHRiMP, mrFAST, mrsFAST, ZOOM, SSAHA2, Mosaik

8. Bowtie

9. Binning reads Need annotated features
Exons, genes, transcripts
For each feature, the total number of reads mapped is produced
Not directly comparable across features/samples yet
Usually followed by normalization

10. Normalizing counts Why normalize? RPKM is most frequently used
Longer features have more reads mapped
Deeper sequencing produces more reads
RPKM is most frequently used
Reads Per Kilobase per Million reads
Defined as C/(LN)
C = number of reads mapped to a feature
L = length of the feature (in kilobases)
N = total number of reads from the sample (in millions)

11. RPKM examples

12. Gene model predicted for fungus Trametes versicolor using
Augustus and RNA-seq hints
Above is a screenshot of Gbrowse instance for fungal species Trametes versicolor for Genozymes project. Project is sequencing both DNA and transcriptome (RNA-seq) and COE is responsible for annotation.
Example of gene predicted using ab intio predictor Augustus (Confident models) using hints from RNA-seq to check accuracy of prediction
- Hints are built from short-read alignment of Illumina RNA-seq spliced reads onto the genome (Mapped Reads)
- Splice reads show direct evidence of introns (next slide)
- Hints are used with ab initio predictors (Augustus) during training and prediction stages

13. Splice Variants

14. “non-coding” RNA molecules
LincRNA-p21
Tran et al.,
In press

15. MIRA Assembly Contig: T_rep_c1201 Read members: 96 Length: 2429 bp
Example
MIRA Assembly Contig: T_rep_c1201 Read members: 96 Length: 2429 bp
Combined Assembly
T_rep_c1201 is part of a 6 member contig
2 are partial transcripts assembled by PTA

16. Detecting Differential Expression
Compare quantification values across samples or across features
Most tools summarize/normalize counts and suggest DE features
Cufflinks/Cuffdiff, R packages (DESeq, edgeR, baySeq, TSPM), SAMtools
DE features go through similar analysis to microarray data analysis (e.g. validation)

17. Cufflinks

18. Cufflinks Tutorial

19. Anaerobic biocorrosion in reactors filled with WP-LS medium

21. SSV1 Replication Cycle (UV Induced)