Error from Longitudinal first-differences

[Shannon_Wilson]

Hi,

Sorry if this has come up before, but after obsessively looking all over the forum, I broke down and made a new post.

I am having a ton of trouble (def user error) running the longitudinal first differences packages. Essentially I am trying to do the same thing this person

How to create a beta diversity plot similar to this one from the Global Gut study?

Hello everyone, In this Youtube lecture by Dr. Dan Knights (https://www.youtube.com/watch?v=M8ylvsS0MHg&t=304s&ab_channel=DanKnights), he mentioned a "neat trick" that they used in the Global Gut study, that is, to plot beta diversity values as single variables by comparing each sample to a baseline/population. [Screen Shot 2021-01-11 at 1.54.40 AM] [Screen Shot 2021-01-11 at 1.55.43 AM] I would like to create a similar plot with my data but I'm not sure I understand this properly and which …

This lead me to try to get first-differences. I coded the following based on the longitudinal tutorial.

qiime longitudinal first-differences
--m-metadata-file corn_meta.tsv
--m-metadata-file shannon_vector.qza
--p-state-column period
--p-metric shannon
--p-individual-id-column participant
--p-replicate-handling random
--o-first-differences shannon-first-diff.qza

I get the following error message:

Plugin error from longitudinal:
'shannon'

Debug info has been saved to /var/folders/ps/vxhcfds57kj3sgnrmh8cr58m0000gn/T/qiime2-q2cli-err-132bwe67.log

A few questions along with this "how do I make the above command run?":

1. the shannon_vector file is directly from my original core-metrics analysis. is this right?
2. individual ID column- does this require to be in the #q2:types? I am trying to examine them based on participant ID number (rather than participant ID and timepoint as my fastq files are ran as). I have tried it both ways and still get stuck on the error described above.
3. do I need to merge my metadata? I have tried the following from the tutorial page
   qiime metadata tabulate
   --m-input-file corn_meta.tsv
   --m-input-file shannon_vector.qza
   --o-visualization tabulated-combined-metadata.qzv

however, when I try to use this metadata, I get the following:

'There was an issue with merging QIIME 2 Metadata:
cannot merge metadata with overalapping columns. The following columns overlap: 'shannon_entropy' '

so based on that error, I am assuming the code will automatically merge the results from my original qza files into my metadata to calculate them, but I am just a little lost on how to get my code to run cleanly!

Any help or advice would be much appreciated

[timanix (Timur Yergaliyev)]

Hi!

@Shannon_Wilson

Plugin error from longitudinal:
'shannon'

Some time ago "shannon" metric column name in the vector was replaced with "shannon_entropy", so I think if you will change it in your command (--p-metric) it should work.

@Shannon_Wilson

the shannon_vector file is directly from my original core-metrics analysis. is this right?

Right!

@Shannon_Wilson

individual ID column- does this require to be in the #q2:types?

Not sure about this one, just try with your current command (with shannon_entropy).

@Shannon_Wilson

do I need to merge my metadata?

No need to do it, since plugin will take care of it.

[Shannon_Wilson]

Perfect! That worked! thanks!

Where can I find a list of the current metric names to use? I am trying to do it on my bray curtis distance and 'distance' isn't working as it is stated in the tutorial.

[timanix (Timur Yergaliyev)]

@Shannon_Wilson

Where can I find a list of the current metric names to use?

For alpha diversity vectors from core-metrics it should be:
pielou_evenness
shannon_entropy
faith_pd
observed_features

@Shannon_Wilson

I am trying to do it on my bray curtis distance and 'distance' isn't working as it is stated in the tutorial.

@Shannon_Wilson

I am trying to do it on my bray curtis distance and 'distance' isn't working as it is stated in the tutorial.

Could you provide a reference to the tutorial you are following, exact command and the error?
I am guessing now but if you are running LME on first-distances output, did you try Distance instead of distance?

[Shannon_Wilson]

Yes, I am sorry. I have been using the longitudinal tutorial page (Performing longitudinal and paired sample comparisons with q2-longitudinal — QIIME 2 2021.11.0 documentation).

I did try both Distance and distance. I was trying however to do a first-difference (the tutorial does it with unweighted-unifrac, so I assumed it would work for all Beta-diversity measures?). Or are there specific ones for each measure like with alpha diversity?

I would also like to create a volatility or some graphical representation of my beta diversity measures. As Beta measures are in a 3D space, I am assuming votaility will not work for that? I remember at a qiime workshop there were moving graphs that did something similar. Where can I find a tutorial on this? I could NOT find one.

Thanks so much for your help

[timanix (Timur Yergaliyev)]

Good news!
@Nicholas_Bokulich answered your questions ! I will copy and paste it here .

@Shannon_Wilson

I did try both Distance and distance. I was trying however to do a first-difference (the tutorial does it with unweighted-unifrac, so I assumed it would work for all Beta-diversity measures?). Or are there specific ones for each measure like with alpha diversity?

@Nicholas_Bokulich

The column name should not change between metrics. But it looks like you are using the wrong command — you would use first-distances to extract first distances from a distance matrix, not first-differences.

@Shannon_Wilson

I would also like to create a volatility or some graphical representation of my beta diversity measures. As Beta measures are in a 3D space, I am assuming votaility will not work for that?

@Nicholas_Bokulich

Yes! First distances will output a vector of pairwise distances between each individual's samples between each timepoint. This can be input to volatility to visualize these distances at each timepoint (the value shown will be each individual's distance between timepoint t and t-1).