Taxa barplot problem with family, order and species

[Linda_Abenaim]

@cherman2

I am gonna repeat some of my questions because I think working through them would benefit you.
What would filtering by prevalence do?
Why do you want to filter by prevalence?
What does setting a prevalence of .34 do?

I’m so sorry, probably I’m confused and I don’t know how answer to these questions because I don’t well understand that means filtered by prevalence.

I understood that is referred to samples but I don’t understand very well the meaning.

I’m sorry! This is my problem

[cherman2 (Chloe Herman)]

Hi @Linda_Abenaim,

@Linda_Abenaim

I’m so sorry, probably I’m confused and I don’t know how answer to these questions because I don’t well understand that means filtered by prevalence.

No problem! Lets take a little step back.

@Linda_Abenaim

I understood that is referred to samples but I don’t understand very well the meaning.

It does have to do with samples, we are on the right track

Lets talk steps for figuring out what prevalence means!

1. Have you check out the docs for qiime feature-table filter-features-conditionally docs? Make sure to read all the parameters and their descriptions!

2. Within the docs for this command, this is how the prevalence parameter is explained: "The minimum portion of samples that a feature must have a relative abundance of at least abundance to be retained." Does this definition make sense to you?

3. Here is a definition of prevalence in the microbiome analysis: https://microbiome.github.io/OMA/microbiome-exploration.html#prevalence

4. Here are some definitions of prevalence that I found online. There are not specifically microbiome related but I think they explain prevalence well. What is Prevalence? - National Institute of Mental Health (NIMH) and Prevalence - Wikipedia

After reading the linked resources, Lets try answering the questions:
What would filtering by prevalence do?
Why do you want to filter by prevalence?
What does setting a prevalence of .34 do?

Hope that helps!

[Linda_Abenaim]

Thank you so much for your precious help like always!

After reading, maybe prevalence is the percentage of samples that have 1% of abundance? If I selected 1% of abundance of course.
So, if I put 1% of abundance and 34% of prevalence this means that I remove the ASVs that are less than 1% of abundance and are present in 34% of samples?

So, if I reduce the percentage of prevalence (probably 5-10%) can I maintain more samples?
Did I well understand ?

If I put 1% of abundance and 10% of prevalence I remove the ASVs less than 1% only in 10% of samples?

[cherman2 (Chloe Herman)]

@Linda_Abenaim

After reading, maybe prevalence is the percentage of samples that have 1% of abundance? If I selected 1% of abundance of course.
So, if I put 1% of abundance and 34% of prevalence this means that I remove the ASVs that are less than 1% of abundance and are present in 34% of samples?

Nailed It .

@Linda_Abenaim

So, if I reduce the percentage of prevalence (probably 5-10%) can I maintain more samples?
Did I well understand ?

You will retain more features. You will lose samples if a sample has no features that meet the criteria, which might have been happening with your first parameters!

@Linda_Abenaim

If I put 1% of abundance and 10% of prevalence I remove the ASVs less than 1% only in 10% of samples?

You are 100% right! Just to rephrase
It will be removed if the ASVs is less then 1% abundance or in less then 10% of samples. Similarly, For a ASV to be retained it has to be equal to/greater than 1% abundance and equal to or more then 10 % of samples.

[Linda_Abenaim]

Thank you very very very much for your help!!

[Linda_Abenaim]

Hi everyone!
I'm here again.

I tried with this command: qiime feature-table filter-features-conditionally
--i-table feature-table.qza
--p-abundance 0.01
--p-prevalence 0.10
--o-filtered-table filtered-table.qza

and this is my visualisation with taxa-bar-plot:
taxa-bar-plots-grouped-clor-mit-conditionally.qzv (416.4 KB)

This was my objective, as @jwdebelius suggested:

@jwdebelius

So, in my opinion, you shouldn't show things iwth less than 1% average relative abundance over your samples because it's hard to read relative abundances below that threshhold.

I saw that some ASVs are 0.8%; is it normal?
Do you think that I'm going the right way?

I also tried to do this:

@cherman2

@Linda_Abenaim

Do you think that I have also to do this command?
qiime feature-table filter-features
--i-table table-no-clor-mit.qza
--p-min-frequency 50
--p-min-samples 4
--o-filtered-table table-filter-abund.qza

This again "Depends"
What are you trying to accomplish? and how do imagine using this command to achieve your goal?

and I saw that there are no differences.
taxa-bar-plots-grouped-conditionally.qzv (421.9 KB)

[cherman2 (Chloe Herman)]

Continuing the discussion from Taxa barplot problem with family, order and species:

Hi @Linda_Abenaim

@Linda_Abenaim

I tried with this command: qiime feature-table filter-features-conditionally
--i-table feature-table.qza
--p-abundance 0.01
--p-prevalence 0.10
--o-filtered-table filtered-table.qza

and this is my visualisation with taxa-bar-plot:
taxa-bar-plots-grouped-clor-mit-conditionally.qzv (416.4 KB)

I agree has a percentage below 1%
d__Bacteria;p__Bacteroidota;c__Bacteroidia;o__Flavobacteriales;f__Flavobacteriaceae;g__Myroides | 0.976%
and your provenance says that you did filter on 1% abundance

Could you send me the table you got as an output for: qiime feature-table filter-features-conditionally?

[Linda_Abenaim]

Hi @cherman2,

this is the table
table-filter-clor-mit-conditionally.qzv (512.2 KB)

Anyway, it is not a big problem for me if I have some percentage at 0.9%, it is only to understand why.

[cherman2 (Chloe Herman)]

Hi @Linda_Abenaim,
I misunderstood how filter conditionally works!

At one point I said:

@cherman2

It will be removed if the ASVs is less then 1% abundance or in less then 10% of samples. Similarly, For a ASV to be retained it has to be equal to/greater than 1% abundance and equal to or more then 10 % of samples.

I was wrong

The prevelance/abundance filter just says its at least 1% in at least 10% of samples. Meaning that if its 1.0% in 10% of samples and 0% in the other 90, that gives a relative abundance of like 0.1% overall. So I believe that you used the command correctly, our understanding of how it filtered was just off!

Hope that clarifies things!

[Linda_Abenaim]

Mmm i’m a little bit confused☹️

So if I want to remove the ASV that are less than 1% of abundance what do you suggest to do?
I want to do this only for the visualization on taxa barplot.

Or maybe it is important also for ANCOM bc..

But for that moment I just want to remove the ASV that are less than 1%.

[cherman2 (Chloe Herman)]

Hi @Linda_Abenaim.
I don't think that QIIME 2 has a way to filter on relative abundance per sample like you want.

[Linda_Abenaim]

Ok.
Is there not a way to group the ASVs less than 1% in "others" for the taxa barplot visualisation?

I saw it in some papers.

[cherman2 (Chloe Herman)]

@Linda_Abenaim,
I am not sure that qiime has that either.

It is a little bit difficult to do grouping like this because if feature 1 in sample 1 had a 20% relative abundance, but feature 1 also had a relative abundance of 1% in sample 2. Would you want feature 1 to be collapsed into other because it has a low abundance in sample 2? I don't think so because you would be missing feature 1's large presence in sample 1. This is what I mean about you wanted to filter on relative abundance per sample.

I would suggest reading the methods of the papers you saw that visualization in and seeing if you can follow their methods! Just make sure to cite them if you do!