(HUMAnN),14 to link the reads to known enzymatic functions and putative pathways.
Wortman echoed what Lita Proctor had emphasized about compositional diversity being greater than functional diversity with respect to variation among different individuals, based on a comparison of phylogenetic analysis and metabolic reconstruction results. In other words, there is a lot more variation in phylotypes than in pathways. “If there is such a thing as a core microbiome,” Wortman said, “it may be at the level of function more than at the level of organism.”
Wortman identified five major challenges to HMP data interpretation:
- Meeting data volume and computational requirements. Reiterating what Proctor said, Wortman urged development of an infrastructure for people to access available data. She also cited a need to increase algorithm efficiency and reduce data redundancy.
- Linking microbiome function to community composition. How can the two different types of analyses (phylogenetic analysis and metabolic reconstruction) be linked so that more nuanced questions can be addressed? (In other words, Which organisms are responsible for which functions?)
- Integrating different types of -omics datasets. All of the shotgun sequencing data are genomic-level data, with very little functional validation that the identified pathways are active. How can genomic data be integrated with transcriptomic, proteomic, and metabolomic data integrated into a systems biology–level approach to studying these communities?
- Modeling microbiome community dynamics. How do microbiomes change over time? What are the drivers of those changes? The environment? Health status? How does the prevalence of certain species affect other species in the community?
- Correlating microbiome shifts with host phenotype. It can be very difficult to associate shifts in community composition (or functional state) with host phenotype when the phenotype in question is not well defined and when the impact of environmental change on that phenotype is unknown.