Metatranscriptomics:
Challenges and Progress
Cindi Hoover
DOE Joint Genome Institute
May 17, 2012
Metatranscriptomics
Metatranscriptome
The complete collection of
transcribed sequences in a microbial
community:
Protein-coding RNA (mRNA)
 Non-coding RNA (rRNA, tRNA,
regulatory RNA, etc)

Metagenome
= who’s there?
Metatranscriptome = function?
What
genes are active in environment?
How does gene expression change in
response to particular conditions?
Evolution of Metatranscriptome Methods
 cDNA clone libraries + Sanger
sequencing (low throughput)
 Microarrays (medium throughput)
 RNA-seq enabled by next-generation
sequencing technologies (high
throughput)
 Influenced by presence of rRNA
Main Challenges
Wet lab
 Low RNA yield from environmental samples
 Instability of RNA
 High rRNA content in total RNA

mRNA = 1-5% of total
http://www.nwfsc.noaa.gov/index.cfm
How do you effectively removal rRNA from
metatranscriptome samples?
Bioinformatics
 General challenges with short reads and large
data size
 Small overlap between metagenome and
metatranscriptome, or complete lack of
metagenome reference
http://cybernetnews.com/vista-recovery-disc/
rRNA Removal Methods
Method
rRNA feature used
Input Manipulate
RNA raw RNA
Before cDNA synthesis
Subtractive hybridization
RNase H digestion
Conserved sequence
High
Exonuclease digestion
5’ monophosphate
Gel extraction
Size
Biased poly(A) tailing
2o structure
Low
Sequence feature
Low
No
High abundance
Low
No
Yes
During cDNA synthesis
Not-so-random primers
After cDNA synthesis
Library normalization w/ DSN
Sample-specific probe method
Stewart et al, ISME J (2010) 4, 896–907

One of the first to successfully
tackle the rRNA in
metatranscriptome problem

PRO: Customized probes are
specific to communities of interest

CONS: Very time consuming
process; requires >3ug RNA or
matched DNA samples
Method has been applied on marine
 Different batches
of probetomay
metatranscriptome
samples
give different
results
substantially
reduce
rRNA.
Epicentre: Ribo-Zero TM
 Essentially a subtractive
hybridization
 rRNA removal reagent
contains oligo probes
complementary to rRNA
sequences
 Magnetic beads bind
rRNA-probe complexes
and remove them from
solution
 Process takes ~1-1.5
hours; requires 1ug total
RNA
Ribo-Zero Types
Metabacteria: handles Gram (-) and
Gram (+)
Human/Mouse/Rat: also works on
fungal samples
Plant Leaf
Plant Seed/Root
Synthetic metatranscriptome
Both methods tested on sample
Mettr_1:
Organism
Amount in pool (ug)
Prochlorococcus marinus
pastoris CMP1986
0.1
Pediococcus pentosaceus
6.0
Acinetobacter sp. ADP1
2.5
Cyanobacterium
synechocystis PCC 6803
3.0
Synechococcus elongates
PCC 7942
0.5
Total Pool
12 ug
Example of Depletion QC
Red = total RNA
Blue = (+) Ribo-Zero A
Green = (+) Ribo-Zero B
Agilent Nano chip: total RNA vs depletion with beta test kit
Initial Mettr_1 Data
Sample
Total
reads
(million)
% rRNA
% Map
Mettr_1 CONTROL
(no depletion)
6.08
75.4
4.1
Mettr_1 (+) probe
6.76
19.3
24.3
Mettr_1 Ribo-zero A
7.96
4.1
68.0
Mettr_1 Ribo-zero B
6.82
4.3
69.5
Gene Expression Correlations Ribo-Zero
vs. No Depletion
No Depletion
0.01
0.1
1
10
100
1000
10000
100000
100000
R² = 0.9999
1000
100
10
1
Ribo-Zero Method
10000
0.1
0.01
Ribo-Zero does not appear cause bias in gene expression.
Gene Expression Ribo-Zero vs Probe
Method
Probe Method
10
100
0.99997
0.99998
1.00000
0.99979
0.99988
0.99980
0.99979
1.00000
1000
10000
R² = 0.9996
100000
100000
10000
1000
100
10
1
0.1
0.01
RiboZero Method
1
CNCB-probe (+)
0.1
CNSZ-RiboZero B5
0.01
Gene Expression Correlations Ribo-Zero
Replicates
RiboZero A
10
100
1000
10000
100000
100000
R² = 1
10000
1000
100
10
1
RiboZero B
1
Ribo-Zero & Cow Rumen
Cow Rumen Data
Sample
% rRNA
% Map (rumen)
% Other
No depletion
control
82.4%
3.4
10.5
Ribo-Zero
Metabacteria
15.9
27.7
55.2
4.9
26.7
56.3
Ribo-Zero
Metabacteria
+
Human/Mouse
/Rat
Ribo-Zero is effective, even on complex metatranscriptome samples
like cow rumen.
Summary
rRNA removal technique is critical to
metatranscriptome sequencing
success!
Ribo-Zero = efficient rRNA removal
method
Highly effective on complex
metatranscriptome samples
Ability to customize by mixing rRNA
removal solutions
Acknowledgements
 Cris Kinross
 Matt Blow
 Jeff Martin
 Weibing Shi
 Shaomei He
 Erika Lindquist
 Feng Chen
Questions?