sequences source

Hi Itai,
how many sequences did you get as the result for your query? what exactly would you like to get as an output for each sequence?
Would you like for each sequence the complete f-score table for all the terms associated with the sequence? Or would you like the best matching term out of a list of possible terms? Or something else?
I think it would help if you could also explain what is your motivation (i.e. why are looking for all sequences of a given class, and what is your end goal with these sequences).

Hi Itai,
There are several ways to approach this, depending on your exact goal:
For a given ASV, we can calculate an f-score quantifying the association for each dbBact term and this ASV. The question is if for each ASV you want the term with the highest f-score, or you want the term with the highest f-score out of a given set of terms.
For example, if you have an ASV that is found in feces of dogs, it could be that "feces" will have a higher f-score than "dog", but maybe you are interested in the animal host rather than the "feces" term.
Or maybe you are interested in something different (for example for each ASV, getting the f-scores for all terms out of a given set, or not using the f-score but a different metric)? Let me know


In any case, the simplest way is to use the calour and dbbact-calour python modules in a jupyter notebook:

assuming you have python (preferably in a conda environemt)

1. install calour (see here)

2. install dbbact-calour (installation instructions are part of the calour installation instructions).

3. use the Experiment.add_terms_to_features() function. See here for documentation

An example notebook can be found here


Let me know how it goes
Amnon