This week I am at the International Society of Microbial Ecology (ISME15) Conference in Seoul, South Korea! I want to share with you some of my favorite talks from each day. Today, I spent most of my time in the “Microbiomes of marine ecosystems” session and made some stops at other sequencing related talks. Perhaps my favorite talk of the day goes to… Antje Boetius! Her talk focused on the microbial diversity of the surface of seafloor sediment and how the microorganisms inhabiting the sediment surface vary from the pelagic community. Boetius showed that there is little similarity between the sediment populations and overlying pelagic zones which, at first thought, is surprising as a large amount of the ocean’s biomass will eventually settle to the sediment surface. However, Boetius reminded us that the cell density of the sediment is around to cells per mL which is a cellular density hundreds to thousands of times more dense than the ocean’s surface and, although there is a large input of biomass from the surface, the sediment population is productive enough to overpower the signal of pelagic DNA. You can check out a paper by Zinger et al (2011) for more information on this topic. In fact, Boetius et al. have followed up the 2011 paper to compare the communities of the pelagic, sediment traps, and sediment surface. What they found was that the pelagic and sediment trap communities are relatively similar while the sediment surface remains distinctive from both the pelagic and trap communities.
The computational tool that Boetius used the most throughout her presentation was non-metric multidimensional scaling (NMDS). NMDS is an ordination technique that will place individuals (samples in Boetius’s case) in an n dimensional space in such a way that the distances d in this [usually] reduced space are as close to the the dissimilarities δ calculated in the pairwise dissimilarity matrix. The pairwise (site by site) dissimilarity matrix Boetius was generated using the Bray-Curtis dissimilarity coefficient of the OTU/taxa tables for all pairs of site. How (what package/program) Zinger et al (2011) perform NMDS on their datasets is not explicitly stated; however, R packages such as vegan can provide you with the means to perform these analyses on your own. With NMDS you have the power to create nice visualizations of your data such as this figure from Zinger et al., 2011:
Since I usually talk about bioinformatics tools, I would like to highlight is CheckM by Donovan Parks et al. CheckM is a tool for assessing the quality and completeness of assembled genomes produced from isolates, metagenomes, and single cell genomes. CheckM makes use of traditional assembly quality assessments such as GC content, coverage, and presence of single-copy genes. The real benefit of this package is that CheckM can identify “genome bins that are likely candidates for merging based on marker set compatibility, similarity in genomic characteristics, and proximity within a reference genome tree.” In Donovan’s presentation, he showed us how an IMG identified contaminated genome was able to be decomposed into 2 complete genomes using CheckM. Unfortunately subterranauts, CheckM depends on comparisons to reference genomes and has not been evaluated for use on bacteria related to candidate phyla. However, a talk by Karen Lloyd alluded to a common ambition in the field of microbial ecology — that single cell genomes, when done right, can serve as a reference genomes for metagenomic studies. For now, this may be a real-life Escher painting as single cell assemblies need to be checked for quality before they can be used for binning of metagenomic sequences and the strongest tools for QC of assemblies are reference dependent… but, despite the difficulty in obtaining of high quality genomes from metagenomes, it appears the community is optimistic.