Is there any method we would say that betadiversity in T1 is higher/lower than the control?

beta
diversity

(Khak Nasheen) #1

Dear Researchers,

I applied pairwise Kruskal - Wallis on alpha diversity indices for my treatments. I got box plots and can deduct the result from the box plot that the alpha diversity in T1 is higher/lower than control. However, I could not get any box plots for beta-diversity, that I may say that beta diversity is higher or lower in T1 than control? I have applied non-parametric permanova and anoism however they just give me a p and q value, which just showed significant differences or no difference. Can anyone help me on this? I want to make box plots that beta diversity in the T1 is higher or lower than control.

Thanks


(Justine) #2

Hi @khaknasheen,

Please check out the Alpha and Beta significance section of the moving pictures tutorial again, and try the --p-pairwise flag on your qiime beta group-significance to get boxplots.

However, it’s worth noting that beta diversity doesn’t function the same way as alpha diversity, since we’re looking at distances (difference) between samples. This has a couple of implications. First, it means that you’re always looking at your data as an intersection between two parameters. Your boxplots are of the distance between group A and itself (within A), group B and itself (within B) and between A and B (a vs b).

We have to make a comparison across the groups, because we can’t develop an innate sense of magnitude. (And sometimes magnitude is contextual anyway.) For instance, if I say something is 2 away, it’s hard to tell what the distance actually is without units. Two could be 2 hours, 2 minutes, 2 leagues, 2 blocks, 2 feet. I And whether that’s “near” or “far” depends on those units. Unfortunately, distances in microbial ecology don’t have easy to use units, so we can’t use that as a measure of “near” and “far” (similar and disimilar). But, if you know the distance from X to Y is 2 and the distance from X to Z is 8, then you know that X is closer to Y than it is to Z.

So, when you’re doing your statistical test, you’re essentially asking if theres a difference in distances within group A or within group B or between A and B. Looking at the boxplots can help you judge if that difference is because there’s a larger within-group distance, or because the between group distance is larger than the within group distance. Each of these, in turn, have different implications for your interpretation…

Best,
Justine


(Khak Nasheen) #3

Great Madam, I understand now
Thanks.