I've been thinking a lot about functional metagenomics recently, and this recent paper from Ohad Manor and Elhanan Borenstein gave me a lot to think about. Microbiome 2017 5:19
DOI: 10.1186/s40168-017-0231-4
When you describe a group of microbes in the human microbiome (let's say the gut), you can (a) write down a list of all of the bacterial and viral species that are present or you can (b) write down a list of all of the genes that are present. Of course, you can do some combination of (a) and (b), or (c) something completely different, but (a) and (b) are good places to start.
The reason I bring this up is that depending on whether you read the literature on (a) species-oriented analysis or (b) gene-oriented analysis, you might get radically different ideas about the microbiome. Recent work on (a) species-level analysis suggests that people generally have distinct, unique sets of microbes in their gut that persist over time. However, work on (b) gene-level (or "functional") analysis suggests that everybody's microbiome is basically the same. Of course, that's an oversimplification, but I think it makes the point.
Getting to Manor & Borenstein 2017, the big point I got from their paper is that one possibility for why we haven't been seeing the same degree of individual uniqueness in functional (or gene-based) profiles of the human microbiome may be because we're not doing the statistical analysis quite right. The figure below shows an example. On the right side in (b) and (c) you see the typical abundance metric in blue, and their improved abundance metric in red. While glycolysis shows a greater range of variation (b), the RNA polymerase pathway shows a much smaller range of variation (c). This makes the important point that the field hasn't yet come to a consensus on how to analyze this data, and improvements in statistical analysis may give us a dramatically different idea for what matters in the microbiome.
There's a lot more detail to the paper, and it's worth a close read if you are working in this field. My personal impression is that everything is still up in the air when it comes to data analysis for functional metagenomics. People come at it from many different angles, all the way from de novo assembly to reference alignment. I'm looking forward to seeing what approaches end up showing us what's interesting and important about the human microbiome.