Consistent with the idea that ASVs categorized as ‘unclassified’ or ‘uncultured’ represent a valuable piece of information. An idea with which I agree. In my analysis of differential abundance at the genus level from 16s rRNA gene sequencing, several of the microorganisms identified as significant are classified as ‘uncultured’ or ‘unclassified’. Taxa unclassified to a given taxonomic rank are named as ‘unclassified’ followed by the highest taxonomic rank to which they could be classified. In line with this idea, it seems coherent to me to perform this same strategy with ‘uncultured’ taxa, i.e. to name these microorganisms as ‘unclassified’ followed by the highest taxonomic rank at which they could be classified. By making this modification there are moderate but considerable changes in my results, what do you think about this? I attach the results by leaving the taxa with the name ‘uncultured’ (followed by an arbitrary number because many genera are found as uncultured) and the results after changing them to the highest taxonomic rank at which they could be classified.
Hello,
Basically, what you are trying to do is the same thing that I always do with such taxa.
However, as you already noticed, you got slightly different results after "masking" such annotation by the higher rank possible and "unclassified".
My guess that by renaming it you lost some resolution since otherwise separated taxa was collapsed into renamed taxonomies you created.
For example, you have 2 taxa that were annotated to the same family but different with different names at genus level (uncultured1 and uncultured2), which were collapsed under one new name "Uncl. Family".
What I usually do is renaming them to "Uncl. Family (uncultured1)" and "Uncl. Family (uncultured2)". It may look a little bit clumsy, but I really prefer to keep such names instead of "uncultured2" and instead of collapsing potentially different genera under one name.
Hope that helps,
Thank you very much for your prompt response. I fully agree with your perspective, and indeed, the collapse of several different ASVs classified as ‘uncultured’ as a single taxonomic unit is a clear artifact of the renaming of the ‘uncultured’ category. In fact, I was looking for a function or code that would automatically rename the ‘uncultured’ ASVs with a consecutive number to discriminate them and not lose resolution. Any idea? or the only solution is to do it manually.
Hello,
I am not aware of any packages that can rename it for you automatically.
So if you still want to rename it you need either to do it manually, or write your own script, or use rename-ids (if you are using collapsed table) to rename them (but you still need to provide a metadata with all renamed IDs, so it is still not automated and require one of the two first options).
Best,
Thank you very much.
Hi @DavidRivera,
You can try using use RESCRIPt's qiime rescript edit-taxonomy ... action to relabel the taxonomy.
Hi, it's me again. I was exploring the options for this task and I realized something. First of all I'm going to highlight some obvious things so that you understand my reasoning. Genus-level ASVs that were classified as ‘uncultured’ are grouped into a single taxa when they are two genera that belong to the same higher taxonomic rank (i.e Family) because the coding I was doing adds the ‘unclassified’ category and the name of the broadest taxonomic range to which it could be classified. Although that collapse could be grouping ASVs that belong to different genera, it could also be grouping ASVs that could effectively be the same genus, we will always have that uncertainty because we don't know the taxonomic classification of the genus. Therefore, at the time of the discussion we could argue that some unclassified genera of family ‘x’ are related to my variable of interest, however, we do not know what specific genera they could be and the thread of argumentation would focus on the general life history of the bacterial family, class or order. What do you think?.
That is correct!
You are right again! I think both possibilities can't be excluded. It is why I prefer to separate them rather than collapsing into one category. In my experience, there are some ASVs within the same genera that demonstrate different behavior, so collapsing them affects their significance.
I think that it is still valid and interesting since other researchers working with the same database may get similar results to yours. So I prefer keeping labels in the format (Family unclassified (uncultured1)), which helps with:
- at least it shows the highest known rank available
- keeps resolution intact
- preserves comparability with other research done with the same database.
Best,
Thank you very much for your time and patience
I did the same treatment to the ASVs classified as ‘Unclassified plus the highest taxonomic rank at which they could be classified’, i.e. I added a consecutive number to differentiate them because, as with the taxa classified as ‘Uncultured’, many ‘Unclassified’ ASVs were grouped artificially into one taxonomic unit, for example, if I had 5 genera of the family Saprospiraceae that could not be classified, they were all artificially grouped together as a single genus ‘Unclassified_Saprospiraceae’, which we are not sure they belong to a single genus.
Therefore, each ASV was differentiated as Unclassified_Saprospiraceae_1, Unclassified_Saprospiraceae_2, Unclassified_Saprospiraceae_3...etc. Additionally, I also noticed that the code I was using to fill in the gaps in taxonomic ranges that were not identified did not cover all of my ASVs, which I filled in manually. Therefore, I recommend manually checking the taxonomy files to make sure that the codes applied have worked correctly.
Again, thank you very much.