1. Transcriptomics
History and practice

2. Early RNA analysis used Northerns: …..One gene at a time
YFG
Label probe
+ hybridise
Tissue sample
Transgenic
Other species
Dwarf WT
Next gene
Quantify
RNA levels
Extract target
RNA

3. Northerns are too slow for Systems Biology where we want to assay ALL transcripts simultaneously
Massive
Datasets
for thousands
of genes
Genes, protein and metabolites link together into biological SYSTEMS

4. Arabidopsis Merged Network 19392 nodes and 72715 edges
Proteins (red) Metabolites (blue) & Genes (green)
19392 nodes and edges
EXAMPLE:
Cytoscape software
Allows the visualisation of all transcript levels for an organism
This one is based on ARRAY data
Arabidopsis transcriptome network
(Ma et al. Genome Research 2007)

5. Post 2000: Microarrays & RNAseq….
Mass transcript profiling: Transcriptomics
Historically (pre-2000):
Sequencing ESTs and ranking representation
Differential display (random 5’ primers fixed polyA primers)
Post 2000: Microarrays & RNAseq….

6. ‘All the genes you want’
Microarrays
Probe preparation
Target preparation
Acquire or
Generate probes
‘All the genes you want’
Extract RNA from your
Control AND your
Experimental plant
Label cDNA
from sample 1 RNA
…and sample 2 RNA
Spot

7. Microarrays Hybridise & Scan Identify ‘spots’ remove background
produce ‘red/green’ ratios
Hybridise & Scan
Link ratio to relative abundance.
Link spot to gene.
Link genes to each other.
Networks / systems

8. Before processing, we have a LOT of spots
‘Landing lights’ xyz normalisation
After processing, we have a LOT of objective data

9. What biological questions can be explored with transcriptomics ?
Learning outcome:
What biological questions can be
explored with transcriptomics ?

10. Arrays can separate similar genes
Pretend specialist microarray. Only 5 genes ALL responding to a hormone:
Plus hormone vs control
(i.e. known / expected challenge)
All ‘on’
The classic types of array experiments:
1. Normal vs challenge (e.g. pathology, induction)
2. Tissue A vs Tissue B
(e.g. muscle vs liver)

11. Remember: Genomes are not tidy – duplication is common
Plant (arabidopsis)
Fungal (yeast)
Animal (human)
This is a big problem for arrays :
Cross - hybridisation

12. Apart from gross syntenic duplication
Gene families (recycling of function) is common:
e.g. in arabidopsis:
Gene family size
Unique 2 3 4 5
>5
35%
12.5% 7%
4.4%
3.6%
37.4%
Proportion of the genome
Conservation at the base-pair level within genes:
37% of genes highly conserved
(TBLASTX E<10-30)
10% partially conserved
(TBLASTX E<10-5)

13. Pioneer arrays were cDNAs
Derived from mRNA amplified by reverse transcriptase and cloned.
Selected based on partial sequence primed from vector cloning sites (e.g. SP6, T7, T3)
Commonly called ESTs (Expressed Sequence Tags)

14. Homologous EST sequence Dissimilar EST sequence
ESTs can be misleading
Gene of interest
Example EST sequence 1
Homologous EST sequence 2
Dissimilar EST sequence 3
On the slide 1 2 3
Labelled target cross hybridises

15. Multiple Short Probes 25-mers
Genechips have better specificity
Known Gene Sequences 5’ 3’
Algorithmic selection
Multiple Short Probes 25-mers
Hybridisation

16. Biotin-labeled transcripts
Example single colour target labelling - 3’ IVT
Fragment
(heat, Mg2+)
Fragmented cRNA B B
Biotin-labeled transcripts
IVT or WT
(Biotin-UTP
Biotin-CTP)
AAAA
RNA
Target Preparation
RNA isolation is the first step. 1-2 hours
The messenger RNA is then reverse transcribed into cDNA (we then go on to make the second strand of cDNA).
4 hours
An in vitro transcription reaction using biotinylated nucleotides is then done to both amplify and label the transcripts hours
These are then fragmented in order to get a more efficient hybridization ( bases pairs is the goal).
1.0 hours
The fragmented target is then hybridized overnight to a GeneChip expression array. 16 hours
When washing and staining of the array is complete it can then be scanned hours
Wash & Stain
cDNA
Scan
Hybridise
(16 hours)

17. Detection: Hybridisation and staining
Array
Biotin labelled
cRNA Target
Hybridisation
Antibody detection

18. Each probe call is derived from the 75% quantile of the pixel values (sweet spot).
All the probes of a probeset (gene) are combined into ONE measure of expression

19. Data handling: Chips need to be normalised against each other.
Each different colour line maps all the intensities of a single chip
They are NOT co-incident lines
(e.g. yellow and black are outliers)
To compare they need to be comparable

20. Average the intensities at each rank
PA PB PC PD PE
Chip 1
Chip 2
Chip 3
Normalisation
Chip 1
Chip 2
Chip 3
Order by ranks
RMA is a very powerful but simple process that works at the probe level
Average the intensities at each rank
Chip 1
Chip 2
Chip 3
PA PB PC PD PE
Chip 1
Chip 2
Chip 3
Reorder by probe

21. RMA Normalisation makes data more comparable
So we can derive / display differentially
expressed genes
..as candidates for further research...
volcano plot
trend graph

22. RNAseq 3 ‘simple’ steps: A complementary solution:
Take an RNA sample, 1. sequence it, align it to the genome. 2. count how many times each transcript appears. 3. work out the frequency of each transcript.

23. RNAseq software – Tuxedo suite (2012)
Bowtie (2009)*: Ultrafast short read alignment
Aligns short DNA reads at 25 million x 35-bp p/h
TopHat: Alignment of short RNA-Seq reads
Aligns RNA-Seq reads to genomes using Bowtie. Identifies splice junctions between exons.
Named after the Burrows-Wheeler transform algorithm (BWT)
Cufflinks (includes cuffmerge, cuffcompare, cuffdiff)
Uses TopHat to assemble the ‘best’ transcriptome.
Estimates relative abundance based on how many reads support each transcript.
CummeRbund: Visualization of RNA-Seq analysis
R package for Cufflinks RNA-Seq output.

24. Fragments Per Kilobase per Million reads
FDR-adjusted p-values (q-values) - replication

25. Mutant
Control

26. Once we have candidates - we can discover their function...
GO annotations of genes higher in muscle
GO annotations of genes higher in liver
Graham et al. (2011) Animal

27. ....and use these to allocate those differential genes
to pathways and biological systems