There were some really cool sessions at ISME today. The morning started off with a plenary talk by Lars Peter Nielsen about cable bacteria and I sampled the sessions on “Unusual strategies of microbial energy acquisition,” “Network microbial ecology,” and “Meta-ome information to microbial ecology.” For each session I have picked a favorite talk starting with Victoria Orphan‘s talk on the archaeal-bacterial partnerships responsible for sulfate-coupled anaerobic oxidation of methane (AOM).
The beginning of Victoria’s talk summarized the work that her group has completed using a combination of stable isotope tracer experiments and fluorescence in situ hybridization coupled to nanoscale secondary ion mass spectrometry (FISH-nanoSIMS). Using these techniques, they are able to obtain single-cell resolution of ANME-SRB aggregates. Her presentation was particularly elegant because, in addition to producing great figures like these (I apologize that they are not from her work..), Victoria’s group was able to 1) identify which microorganisms are bacteria vs archaea in an aggregate and 2) identify which cells are the most active. She then took this work one step further by using geostatistical methods to identify “hot spots” in the aggregates. The details of this technique where omitted in the presentation but, from this black-box technique, she was able to conclude that, at the individual cell level, the spatial proximity of archaeal and bacterial cells was not well correlated to the hot spots of elevated cellular activity. These finding are inconsistent with previous diffusion-based models proposed by Orcutt and Meile (2008) and, therefore, Victoria finished her talk with the framework for a new model of ANME-SRB interaction in AOM aggregates.
In the “Network microbial ecology” session, I would have to pick Jed Furhman‘s presentation of network analysis from the San Pedro Ocean Time Series. I may be biased as he is a fellow Trojan, but this dataset (collected every month since 1998) is truly phenomenal. With a lot of data to play with, Jed performed local similarity analysis on molecular and environmental time series data and, subsequently, visualized these correlations using networks. Kudos go out to the group who (from the talks I have attended) are the only group to discuss protists, bacteria, and viruses in the same session! I would recommend this paper to all interested in learning more about SPOT and Jed’s work. From a bioinformatic standpoint, Jed continually stressed that these networks are more of exploratory tool than definitive truth.
Finally, the bioinformatic tool of the day and favorite talk in “Meta-ome information to microbial ecology” goes to Mitch Sogin. He presented Oligotyping and Minimal Entropy Decomposition (MED, joint work with Murat Eren and others). Oligotyping is a supervised method that uses Shannon entropy to separate groups of closely related sequences into oligotypes. Thus far, oligotyping has been used on a variety of datasets (Eren et al., 2011; Eren et al., 2014; Maignien et al., 2014). MED is the natural successor to Oligotyping. The details of MED are awaiting publication but Mitch described it as an unsupervised algorithm to separate sequences into groups based on entropy. The aim of MED is similar to OTU clustering in the sense that you want to group sequences derived from the same organism. Now, what makes MED special is that, like Oligotyping, it is able to tease apart oligotypes that may otherwise be merged during OTU clustering but, unlike Oligotyping, it can be applied to a dataset composed of different taxa whereas Oligotyping is designed to analyze only 1 group of sequences (OTU/species) at a time. We were given a taste to some of the results of MED on real datasets and it looks very promising. Keep an eye out for this paper!