# Adonis and ANCOM with rare features **URL:** https://forum.qiime2.org/t/adonis-and-ancom-with-rare-features/17803 **Category:** User Support **Tags:** diversity, composition, ancom **Created:** [December 15, 2020, 5:59pm UTC](https://forum.qiime2.org/t/adonis-and-ancom-with-rare-features/17803 "2020-12-15T17:59:42Z") **Posts on this page:** 1 **Showing post:** 4
### Author: ![jwdebelius](https://forum.qiime2.org/user_avatar/forum.qiime2.org/jwdebelius/32/9655_2.png) [@jwdebelius](https://forum.qiime2.org/u/jwdebelius) #### Post date: [December 16, 2020, 11:13pm UTC](https://forum.qiime2.org/t/adonis-and-ancom-with-rare-features/17803/4 "2020-12-16T23:13:49Z")
Hi @hsapers, I'm going to disrupt some of the order and jump around. > [@hsapers](#): > > I think there is something wrong with my meta data - it will take me a couple of days to comb through it. I’ll make a new post if the error persists. Metadata is one of the hardest parts of any microbiome project. I swear, I have multiple projects where I've spent more time and effort formatting metadata than I have actually doing analysis. [Keemei](https://keemei.qiime2.org/) may help some, it's a good first check. If you've already validated there, I would suggest moving into R or python with pandas to do a more in-depth check. I think metadata validation might be a good discussion as a separate topic. > [@hsapers](#): > > Yes - my sample size is quite low - it’s a very difficult environmental system to access and it just isn’t possible to increase sample size other than to keep going back and use different time points as replicates. The problem is the source community isn’t stable enough to assume these are replicates. Unless I consider that a between sample comparison? If I treated each time point like a different subject in a gut microbiome experiment - it’s the same source community, but variation over time leads to slight differences - perhaps analogous on how there are slight differences in the starting gut microbiomes of different subjects? > > This is the basic set up (with slight variations in design between deployments because something _always_ breaks in the field): > > Slow: 2x glass substrate; 2x crush substrate (all containers plastic) > Fast: 2x glass substrate (glass container); 2x crush substrate (glass container); 2x crush substrate (plastic-lined glass container) > > After indicating that there is not a significance difference between 2x crush substrate (glass container) and 2x crush substrate (plastic-lined glass container) in the fast condition, these were pooled into the same conditions such that: > > Slow: 2x glass substrate; 2x crush substrate (container material irrelevant) > Fast: 2x glass substrate; 4x crush substrate (container material irrelevant) Thank you for the description! It really helps to visualize the design. Although I think one caveat to consider in your combination if you used a single time point is that your permutative p-value is bounded by the number of possible permutations and there are 4 ways to permute 4 samples, giving you 24 possible permutations, for a minimum p-value of 0.04. > [@hsapers](#): > > **2a** Would using `permdisp` address over sampling in one group? I’m going to put a pin in the ‘permdisp’ error while I comb through my metadata - but this sounds like a potential way to evaluate unbalanced experimental design? As I conceptualize the design and try to find an analogy that's maybe easier to find a model for, analyzes that work with nested animal designs might help you. So, for rodents, we have to account for a cage effect because mice are coprophagic little bastards (and rats are coprophagic sweethearts). So, as you have 2-4 independent environmental systems in each group, and within each system, you have 5 nested timepoints. So, you need modeling that lets you account for repeated measurements in your system or some way to nest the data. Permdisp has some issues with severely unbalanced designs - like 3 - 5 times or more samples in one group than the other; but it will help. Although, having seen your design, I think the unbalanced aspect is less of a problem than your number of samples. > [@hsapers](#): > > `fill + container` calculate effect of levels of the factor fill regardless of the level in factor container (ie across all levels of container), and then out of the residual calculate the effect of the levels of the factor container (completely independent of the factor fill). `container + fill` is the inverse and why order matters here. Yes, that's correct. > [@hsapers](#): > > I’m also having a bit a difficulty conceptualizing how the equation in Anderson 2001 is implemented in adonis. According to the [R formulae page](https://rdrr.io/r/stats/formula.html) > > A∗BA\*B is interpreted as A+B+A:BA+B+A:B > > where AA is the main effect of factor A and BB is the main effect of factor B and A:BA:B is the interaction term. > > Wouldn’t that make `A*B` different than `B*C` - and I can’t reconcile that with the equations on page 39-40 of the Anderson paper. Or in this case is the interpreted equation ( A+B+A:BA+B+A:B ) indicating each of the calculated effects - the main effect of A, the main effect of B, and the interaction and not how each of those effects are calculated? I think if you want to get the deep implementation, it can be helpful to go read the code? From my practical perspective i tend to inter the equation as (A + B + A:B) (so, A + B + interaction) and move on from there. That's somewhat aligned with what you see in Figure 7 (page 40), at least based on my reading. > [@hsapers](#): > > I’m not clear on how the slash in interpreted here: > `fill + container/replicate` is this the same as using `fill + container` for each replicate individually rather than pooling all replicates from the same condition? Wouldn’t there only be one sample in each condition then? In R, a `/` indicates a nested variable. So, you can nest blocks based on your treatment. I may have mis interpreted your replicate - I was assuming technical replicates from the same site/time point verses multiple parallel systems the same timepoint. (So, I assume the replicate here was dependent verses independent.) If you have multiple nested factors (like mice in a cage), nesting can help address some of the variance there. So, this may be less relevant given more information about your design. Coming back to 4... > [@hsapers](#): > > **4** Yes - my sample size is quite low - it’s a very difficult environmental system to access and it just isn’t possible to increase sample size other than to keep going back and use different time points as replicates. At your current sample size, or my understanding of your current sample size, changing the pseudocount and transformation isn't going to solve your core problem, which is that compared to your multiple correction penalty, your sample size is too small. I can give you lots of studies where they can't detect individual features as differences (and plenty more where I dont trust the features they found because of their sample size and methodology). If you want to take your time point into difference abundance, you might look at this recent thread about building nested models using LMEs (accounting for your temporal variation). My recent experience with this is that they also tend to be relatively low power and you may not detect anything significantly different. > [@Songbird on paired samples](https://forum.qiime2.org/t/songbird-on-paired-samples/14379): > > Hi all, especially @mortonjt , I'm trying to figure out Songbird on paired samples. Ive been through the tutorial and looked at the differential code from the original paper. Im slightly confused how to use this for paired samples, since as far as I can tell by the formula, the code is looking at a bulk comparison between before and after tooth brushing without conditioning on the individual. (The code is C(brushing\_event) and I might have expected something closer to the (C(brushing\_event) | h… I think, in general, microbiome differential abundance has swung more conservative. Best, Justine --- _[View the full topic](https://forum.qiime2.org/t/adonis-and-ancom-with-rare-features/17803)._