Executive Summary Of External Panel Report

An External Panel convened on February 3, 2012 to assess the NIH Human Microbiome Project (HMP) and concepts under development for a new human microbiome research program. External Panelists included:

Julian Davies (HMP External Scientific Consultant; University of British Columbia)
Francis Ouellette (HMP External Scientific Consultant; Ontario Institute for Cancer Research)
Richard Blumberg (HMP External Scientific Consultant; Brigham and Women's Hospital, Harvard Medical School)
Robert Holt (HMP External Scientific Consultant; Canada’s Michael Smith Genome Sciences Centre)
Alan Williamson (HMP External Scientific Consultant)
David Relman (Stanford University)
Margaret McFall-Ngai (University of Wisconsin, Madison)
Michael Fischbach (University of California, San Francisco)
Ruth Ley (Cornell University)

As part of the HMP assessment, panelists received information from investigators in each of HMP’s six initiatives on the successes and challenges for the initiative, as well as about the lessons learned from the HMP and the path forward for microbiome research. Panelists also had the opportunity to ask questions of the HMP investigators to guide their determination of the significance and impact of the program. The panel identified the following as the key products and resources from the HMP:

A collection of nonpathogenic (reference) microbial isolates and their genome sequences
Longitudinal reference data sets of microbiome composition across the human body.
Computational tools for analysis, interpretation and visualization of metagenomic data.
Clinical protocols for microbiome sampling across 18 sites of the human body.
Bioinformatic protocols for processing microbiome data, particularly taxonomic and functional metagenomic sequence data.
Establishment of a Data Analysis and Coordinating Centre (DACC) that enabled the dissemination of HMP resources to the scientific community.

There was consensus that the HMP raised the visibility of the nascent field of human microbiome research on a global scale and brought together a diverse community of scientists from a broad range of disciplines. This was possible because of the trans-NIH Common Fund mechanism used to support the program. The publicly available, quality controlled datasets, tools and standard operating procedures created by the HMP are being leveraged by researchers in the larger scientific community to produce their own high-quality microbiome data that can be compared to HMP and other microbiome datasets.

Lessons that were learned during the course of the HMP and should be taken into consideration in future human microbiome research programs include:

Defining a healthy microbiome is and will be more difficult than anticipated.
Existing tools for processing, analysis, and interpretation of microbiome data need to be improved to increase their robustness to handle increasingly large volumes of complex data.
An effort needs to be made to integrate bioinformatics expertise into clinical research projects right from the onset of these projects through multidisciplinary teams.
Management of the program needs to be established with a clear organizational and decision-making structure from the beginning of the program. A leader outside of the funding agency is needed to oversee the progress of any program.
The need and usefulness of an external advisory body, such as currently exists in the External Scientific Consultants, should be well-defined at the beginning of any program.
Broad consents should be obtained for microbiome research that take human signal “contamination” into consideration with regards to data storage and release.

The HMP has met its original goals of understanding the composition of the communities of microorganisms that live on and in the healthy human body. Before the role of the microbiome in disease can be elucidated, the key functional properties of a normal host-microbiome system and how it contributes to health must be understood. This understanding of the microbiome was beyond the scope of the HMP. A new microbiome program should address this gap. Since the microbiome does not function in isolation, the role of host properties in maintaining a healthy microbiome should also be considered. It is essential to support studies that explore the interactions between the host and the microbiome (including eukaryotic microbes and viruses), with appropriate animal models being developed to explore the full extent of this interplay. Studies of the communication between the host and the microbiome (the metabolomics space) will require more emphasis on the use of advances in modern instrumental technology. Bioinformatics infrastructure and computational tools need to be strengthened and developed for systems level analyses of DNA, RNA, proteins and small molecules as well as for the visualization of functional properties. Capacity must be built for large-scale, high throughput functional assays for the microbiome at the community and strain levels and for host tissues with particular attention to immune system properties. Model systems and related facilities need to be developed, including nontraditional and simpler animal models and gnotobiotic facilities. The panel believed that the following topics were of particular importance to address:

Small molecules from the microbiota (metabolomics, etc).
Host systems including the immune and inflammatory system, the central and enteric nervous system, and the enteroendocrine system.
Human subjects as well as a broad range of model systems of varying complexity: germ- free and conventional mice, zebrafish, synthetic tissue models, cell lines.
Spatial characterization of intact microbiomes (e.g., imaging mass spectrometry).
Baseline signatures from remote/isolated populations (e.g., hunter-gatherer) while they are still largely isolated.

The goals of a Microbiome 2 Program are too extensive to be supported by one NIH institute. The program is an ideal fit for the Common Fund criteria. NIH should also consider the possibility of the formation of a Microbiome Institute.