My presentation at the 2014 MetaCenter symposium in Eugene, OR entitled "High-throughput annotation of metagenomes reveals community physiological variation in the mammalian microbiome"
1. High-throughput Annotation of
Metagenomes Reveals
Community Physiological Variation
in the Mammalian Microbiome
Thomas J. Sharpton
Oregon State University
@tjsharpton
2. How do Microbial Communities
Interact with their Hosts?
• Microbiome structural changes often associate
with changes in host physiology
• Knowing the mechanisms of interaction between
microbiome and host physiology may yield
improved diagnostics or therapeutics
• Metagenomic functional annotation can reveal
how microbiome physiology scales with host
physiology
5. Statistical Simulations Assess Accuracy
and Properties of Annotation
• Simulated mock communities and
metagenomes
• Predicted community physiology
using Shotmap
• Compared predictions to mock-
community truth using Bray-
Curtis distances (L1)
Stephen Nayfach
Pollard Lab
Poster #02
9. Community Structure Varies Across
Inflammatory Bowel Diseased Patients
Adapted from
Qin et al Nature (2009)
10. Crohn’s Diseased Patient Microbiomes
Exhibit Physiological Differentiation
Adapted from
Qin et al Nature (2009)
11. Physiological Alpha-Diversity is Lower
in Crohn’s Diseased Patient
Microbiomes
R
Richness Shannon Entropy
*
*
Crohn’s
Disease
Healthy Ulcerative
Colitis
Crohn’s
Disease
Healthy Ulcerative
Colitis
* p < 0.05
12. Specific KOs Differentiate Crohn’s
Diseased Microbiomes
* *
Crohn’s
Disease
Healthy Ulcerative
Colitis
Crohn’s
Disease
Healthy Ulcerative
Colitis
Abundance
K05595 - multiple antibiotic
resistance protein
K09684 - purine catabolism
regulatory protein
* q < 0.2See Morgan et al. Genome Biology (2012)
13. IBD-Associated Longitudinal Variation
in Microbiome Physiology
• TGF-B-DNR mice
Develop an IBD
akin to Crohn’s disease
• Generated Metagenomes
Before and After Disease
Onset
15. Conclusions
• Shotmap is a new tool that can quantify
physiological variation between metagenomes
• Metagenome annotation accuracy depends on
the parameters selected and data properties
• Microbiome physiology varies in association
with inflammatory bowel disease state
16. Acknowledgements
Collaborators
Stephen Nayfach (Gladstone), Poster 02
Katherine Pollard (Gladstone)
Patrick Bradley (Gladstone)
Tim Laurent (Solozyme)
Stacia Wyman (Gladstone)
Shomyseh Sanjabi (Gladstone)
Jonathan Eisen (UC Davis)
Shotmap is Open Source. Download here:
www.github.com/sharpton/
Funding
Gordon & Betty
Moore Foundation
NSF/NIGMS Joint
Program in
Mathematical Biology
#DMS1069303
NIH R21 AI08953
Developers of open source software
and providers of public data
21. Metagenome Annotation
• Classify metagenomic sequences into functionally annotated
protein families
• Several great tools currently exist
• But, few provide extensive analytical control and force users to
adopt settings that may not be appropriate for their analysis
– Use specific databases or annotation paradigms
• Additionally, many require the user to upload data to the cloud,
which may yield infrastructural bottlenecks as data volume
increases
• We sought to develop a high-throughput workflow that provides
users analytical flexibility, can be run on a multi-core machine in a
laboratory, and handle the entire annotation process, from raw
metagenomic data to statistical comparisons across samples
Editor's Notes
It is less well-understood how microbiome physiology associates with such changes in the host.
There are many methods by which these mechanisms can be elucidated. One that has been very powerful is:
By evaluating how microbiome physiology scales across host physiology, we can develop testable hypotheses about potential mechanisms of interaction between microbiomes and their hosts
Shotmap is:
Analytically flexible and allows users to tune workflow parameters and conduct analyses appropriate to the data being investigated.
For example, it is agnostic to the search database
Extensible. The publically available software accommodates a variety of search algorithms, including blast and fastblast tools as well as profile-based alignment tools like HMMER
Infrastructurally flexible. It can either be run on a multicore lab server or it can interface with a distributed computing environment. It can also optionally communicate with a relational database server to handle the management of results.
I’m not showing the data here, but we see the same pattern with beta-diversity
A virulence factor in several other disease systems, may trigger inflammation
Purine catabolism part of a pathway found to be significantly elevated in healthy patients in morgan et al
Mounting evidence indicates that flagellen may trigger immune activation via innate and adaptive immune systems.
Elson lab shown that flagellated bacteria are implicated in Crohn’s disease
Involved in Vitamin D biosynthesis; Vit D
Note that there are other excellent tools available to researchers in the community
Note Morgan et al study, which explored distinct patient population
Note that many of the KOs found in mice are also found in human clinical study, indicating that there may be consistent physiological differences across species