Using adonis and betadisper from the vegan R package to compare groups (CC208)

[@ddelanira]

This video saved me a lot of time and headache, thank you so much for providing a complete explanation of adonis (usage and correct interpretation)! Your tips on tidyverse are the best, and incredibly helpful! Thank you!

[@Riffomonas]

My pleasure! Thanks for watching

[@analopez3799]

Congratulations on these amazing tutorials! It's a complicated matter for someone with little experience in R, but you make it so much easier to understand. I'm working on my doctoral dissertation at the moment and this helped me a lot. THANKS!

[@Riffomonas]

My pleasure Ana! Thanks for watching 🤓

[@miissJoceLyn]

This is an outstanding explanation of everything you are doing here. This is amazing, this content really contributes to science. Thank you so much

[@Riffomonas]

my pleasure! thanks for watching🤓

[@sarahb4270]

i really appreciate the time and energy it took to make this video. i found it very helpful. thank you!!

[@AS-xf6iu]

I have been subscribing to your videos over several months now. Thank you SO MUCH! These are incredibly informative, helpful, and approachable.

[@Riffomonas]

You are so welcome! Thanks for watching 🤓

[@alba-2460]

These videos have helped me a lot in my analysis, thank you.

[@patschloss3342]

My pleasure! Thanks for watching Alba 🤓

[@Riffomonas]

🤓

[@d0onut]

You're a Saint for this one, thanks for explaining

[@Riffomonas]

Thanks for watching!🤓

[@valpal2424]

Thank you so much! the explanation on betadisper helped me a lot with my uni paper, to make sense of what I even did :)

[@Riffomonas]

Ha! Thanks for watching 🤓

[@nguyenlephuong9489]

Thank you very much for your lecture. I would like to update, from R version 4.1 with the recent vegan package, the command adonis needs to be changed to adonis2() for running without the errors: Error in `colnames

[@Riffomonas]

Thanks I was just noticing that this week

[@sven9r]

very nice episode :) When working with ordinations - I always add the stress or eigenvalues on the axes: With something like that: NMDSstress

[@alvarracles]

Thanks so much for the example. Really healpful, this information is not aviable easilly and free!!!

[@gimanibe]

Very cool, thanks Pat - hard to find applications of adonis online. Since the test is sequential, maybe I would also run a model with factor variables inverted in their position.

[@Riffomonas]

Thanks for watching! 🤓

[@katimbaue]

This is GOLD. Thank you

[@MrSpd85]

I love your videos! Have you considered to make a new one about Poisson GLMs to analyze the abundance? I think it is very necessary! best!

[@Riffomonas]

Thanks! Maybe we’ll revisit that in a future episode but I’m done with microbiome stuff for now

[@hosseinkarimi1981]

Thanks very much, Pat. I don’t know why my AMOVA results in Mothur are different from Adonis results in Vegan package in R.

[@Rydaholic]

Thanks Pat! Really helpful and important video for me! Are you limited to a single "random effect" with the strata function? Or can you add more? For example if your mice came from different litters. Can you account for repeated measures on animal and a random effect of litter at the same time?

[@Riffomonas]

To be honest I’m not really sure. Sorry!

[@thorbjornf.8873]

If you find a solution, please let me know :)

[@christian-1629]

Thank you so much, Pat! This is a fantastic video and incredibly useful for my doctoral dissertation. I do have a stupid question: what was the reasoning behind filtering 1828? Is it possible to analyze the dataset without applying the filter?

[@bellatsachidou3992]

I have a question... If the betadisper is significant does that mean that the significant adonis results are not credible? If you have a significant betadisper do you even proceed to adonis? Thanks for all the tutorials! They are really helpful!

[@Riffomonas]

I haven't seen a problem with a significant betadisper and have a previous study that doesn't see the effect that others worry about www.nature.com/articles/ismej20085

[@YangPeng-vh1nu]

Sorry, I am a little confused about the command 'bd

[@svetlinavasileva109]

Hey Pat, thanks once again! I was wondering what is your approach to subsets. If I want to run a secondary analysis on just a part of the whole sample, is it better to create a new distance matrix for the subset only or is extracting the samples from the full matrix better?

[@Riffomonas]

The pair wise distances won’t be affected by what other samples are included so it shouldn’t matter

[@outsandabouts]

Great video, I just have one question: is there a particular case in which one of the centroids could be outside the ellipse? This may be explaining why there are significant differences between both of my groups but in the plot the ellipses overlap a lot.

[@Riffomonas]

I don’t think it would be possible for a centroid to be outside of its own ellipse. You could certainly have non overlapping ellipses

[@BrennaEgan]

Is it possible to run a post hoc test on the results from adonis? I am testing differences in mean between 6 groups. Adonis was significant for location (betadisper was not significant), but I don't know how to test differences between the 6 groups. I've seen some documentation about using TukeyHSD as a wrapper on betadisper for pairwise differences, but I'm unsure how to implement this. Thanks!

[@Riffomonas]

Aside from the wrapper, you could subset for the different pairs of groups and then correct the pvalues for multiple comparisons

[@shannonbuttimer96]

Since you're giving betadisper a distance matrix instead of an ordination and the method it uses is PCoA, can we actually relate this to the NMDS plot? Is there a way to tell betadisper to calculate NMDS axes to use to calculate distances to the centroids?

[@Riffomonas]

Thanks for watching Shannon! I think if the ordination as a way of viewing the distribution of the samples, albeit in 2D. Betadisper is calculating the distance to the true centroids of the multidimensional rather than 2D space

[@shannonbuttimer96]

@@Riffomonas that makes sense! thank you so much for making these videos!

[@asdmdomf]

Thank you! Does a non significant betadisper-permutest mean that there are no significant differences in variance between two groups? Do you also can see that there is no difference in Variance visually in your PcOA_plot? I`ve created an PcOA-Plot that shows clear difference in Variation between two group. But betadisper-permutest on that data gives no significant result

[@ΑγλαΐαΣκοτίδα]

thank you for this useful video! when I try to run the metaMDS with scores I get an error message for wrong dimensions. any possible solution?

[@Riffomonas]

Hmm. I’m sorry but I don’t really know.

[@svetlinavasileva109]

stackoverflow.com/questions/72483924/species-scores-not-appearing-from-metamds , here is a solution :)

[@Sushinugget]

But now strata is not working!and giving error

[@abhisheknaik7506]

You rock, thanks for this

[@Riffomonas]

My pleasure- thanks for watching!

[@60Aves]

Super useful, thank you!

[@grahamsharpe9812]

Why do you use tibble? And not modify data in excel? Or is this a preference thing?

[@Riffomonas]

It's about reproducibility and transparency. Modifying data in excel is very much not reproducible or transparent. It's next to impossible to document changes in excel files. Also, excel is $$$ whereas R is free. With R, we can document all of the changes in the script and rerun the script multiple times without having to worry about breaking things. Also, if I make a mistake, I can easily correct it by reprocessing the raw data with the corrected code.

[@chitrarajagopal8434]

Sir, thank you for your informative video. When I am performing a betadisper, it gives an f value less than 1..will that be a problem?

[@Riffomonas]

That likely means there's no significant difference.

[@silulekomkhize7839]

Is adonis the same as running a PERMANOVA test?

[@Riffomonas]

yep! it's a tool with many names - adonis, adonis2, permanova, amova... 🤓

[@1234567890sunshine]

Geez I was so lost before you even started the analysis

[@Riffomonas]

Sorry!

[@annamalai993]

Is there any direct link for the data...?

[@Riffomonas]

Try this... github.com/riffomonas/distances/tree/347ad108266734d6ef2de28345b1e75942fc7e68/data

[@franklingarciafernandez5387]

Thanks

[@Riffomonas]

My pleasure Franklin!

[@MattChrisWedding]

You’re great. Perhaps consider lowering your caffeine intake Once I got a few basics your lesson taught many tricks

[@Riffomonas]

lol - reduce the caffeine? NEVER! 😂 thanks for watching