I am working with data from a longitudinal study with 2 time points (T0 and T5).
I applied the q2-longitudinal plugin. I performed the beta diversity and I used the “qiime longitudinal pairwise-differences” command to investigate the differences between T0 and T5 within a biological group for each axis of the PCoA.
Only the axis 2 (which explains the 16.15% of the variation) was able to discriminate my samples T0 vs T5. What does it mean?
This effectively means that samples moved across PC2 in the same direction between T0 and T5, suggesting that these samples experienced a shift in whatever differences in community composition are associated with that axis.
This is useful because T0 and T5 samples may not cluster along that axis, but individual paired samples experience the same directional shift. So compositions are changing in the same way, though not necessarily to the same magnitude or resulting in the same final composition.
This is useful for human studies and other biologically noisy situations where baseline and final microbiota composition are largely shaped by inter-individual variation, causing samples from the same subject to cluster together on a PCoA plot (i.e., all samples from that same patient most closely resemble one another). This inter-individual variation usually still overwhelms any treatment effect, but treatments can cause a directional shift from baseline, indicating a uniform effect on composition. Even if T5 subjects do not cluster together along an axis, shifting in the same direction along this axis can be an important indicator of treatment effect. Exactly what compositional changes that is associated with is another story...