I've seen many papers use PCoA plots/beta diversity analysis to describe differences or variation in microbiome structure, usually something along the lines of "the structure was associated with sampling site" or "microbiome structure varied according to treatments". But I am struggling to understand the core definition of microbiome structure, and how one would describe it. Can someone point me in the right direction or let me know what papers I should read? Thanks!
Metagenomics analysis or 16S analysis ends up in most case on the OTU table. OTU table is a double entry table where row are bugs and columns are samples. For instance, after sequencing 4 differents samples, you end up with the following OTU table .
| | sample1 | sample2 | sample3 | sample4 | |----------------|---------|---------|---------|---------| | Staphylococcus | 43 | 32 | 2 | 0 | | Escherichia | 1 | 5 | 234 | 234 | | Salmonella | 2 | 5 | 63 | 54 |
It reads like this : Sample1 has 43 reads assigned to Staphylococcus.
Alpha diversity describe the content of one sample with different metrics.
For instance, the "richness" is the number of different species in one sample . (e.g Sample1 has three species and sample 4 has two). There are more metrics like evenness, Chao1, Shannon, ACE ... @See Here
Beta diversity evaluate the whole content of the OTU table. In most of the case it does a multivariate analysis.
For example, with the OTU table above you can draw a 3d plot with 3 axes ( Staphylococcus content, Escherichia content, Salmonella content) and draw each sample as a point with abundance of reads. The coordinate of point sample1 will be ( x=43, y=1, z=2). you will quickly see that the close points have similar microbiota. For instance Sample1,2 points are close to each other meaning a similar microbiota structure. In real life, you have more than 3 species, meaning more than 3 axis which you cannot draw. So you must reduce the dimension and create 2 or 3 virtual axis. This is called "ordination". There are several methods. The most famous is the Principal Coordinates analysis which compute distance between sample (point) using different metrics like euclidian distance or BrayCurtis distance. After plotting your ordination, you may see cluster of point which may correspond to something. For example, you can color point by their body source. Sample 1,2 comes from mouth while sample 3,4 comes from bowel.
Try to understand the following beta diversity analysis coming from 131 samples from differents body site. Structure, function and diversity of the healthy human microbiome