Develop #26
Develop #26
Conversation
distribute library with pip
spelling error
…into Biomerene-patch-1
Biomerene patch 1
…complete as default medium name
hariszaf
left a comment
hariszaf
left a comment
There was a problem hiding this comment.
Thanks @Biomerene !
Very useful PR!
I have a couple of minor comments.
Once we resolve those, we could merge the PR to main and then check what if anything from #19 is still needed
| elif modeltype: | ||
| if modeltype == 'ModelSEED': | ||
| self.path = TRAINED_NN_MSEED | ||
| elif modeltype == 'BiGG': |
There was a problem hiding this comment.
@Biomerene do we know support BiGG all the way through? 😃
There was a problem hiding this comment.
Gap-filling BiGG models should work now yes, the models do not get refined based on the aliases (which are part of the ModelSEED database). See also my comment on the refinement in the model build (#26 (comment)).
dnngior/NN_Trainer.py
Outdated
| return o | ||
|
|
||
| def generate_training_set(data,nuplo, min_con, max_con, min_for, max_for, del_p, con_p): | ||
| def generate_feature(data,nuplo, min_con, max_con, min_for, max_for, del_p, con_p): |
There was a problem hiding this comment.
add a space after data,.
dnngior/__init__.py
Outdated
|
|
||
| import os | ||
| os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3' | ||
| os.environ['TF_CPP_MIN_LOG_LEVEL'] = '5' |
There was a problem hiding this comment.
Is tensorflow now a strong dependency ? if it is optional, then i think this may be better to be under an if..else statement because I guess it would lead to an error.
There was a problem hiding this comment.
I moved this documentation to be loaded when the tensorflow package is loaded (442d04b)
| # 1) Add metabolite info | ||
|
|
||
| if metabName in compounds_aliases_dict: | ||
| metab.annotation = compounds_aliases_dict[metabName].copy() |
There was a problem hiding this comment.
I guess this is related to #24 right?
If yes, then it would be a good thing to edit your PR message and mention that the issue is addressed in this PR, to be able to track it! 😉
There was a problem hiding this comment.
It is related, I think, but I am not entirely sure if this issue is entirely resolved now, we would need to retest the models with memote but I have not done this
|
|
||
| reac.name = reac.id | ||
| elif dbType=='BiGG': | ||
| print("skipping refinement, not currently supported for BiGG models") |
There was a problem hiding this comment.
Hmm..
I am a bit confused with the BiGG case.
Would you like to make clear what parts of the dnngior workflow are supported for BiGG models and what are not, so we implement those in the fututre?
There was a problem hiding this comment.
If I remember correctly, this function makes use of some additional information and functions from ModelSEED biochemistry to improve the annotations, I think @danielriosgarza build this part. This is not there for the BiGG database so right now this step is just skipped. I am not sure if we mention the refinement of the ModelSEED models.
| else: | ||
| sys.exit('# ERROR: your output_folder should have parents') | ||
|
|
||
| def main(): |
There was a problem hiding this comment.
Since this is your main function for the CLI, I suggest you add some documentation
There was a problem hiding this comment.
I added some documentation (with some help from collab) see commit c98534a
dnngior/fasta2model_CLI.py
Outdated
| gapfill_model_wrapper(args) | ||
| print('# Done') | ||
| else: | ||
| sys.exit('I dont think this message can show, if it does, you (or more likely me) did something weird') |
There was a problem hiding this comment.
Consider editing this message returning something like
"either a fasta_folder or a model_folder needs to be provided"
There was a problem hiding this comment.
This should actually be already be caught by the argument parser, but this is more informative in case that did not work correctly, I updated it.
dnngior/reaction_class.py
Outdated
| side = -1 | ||
| else: | ||
| mets[i] = stoc | ||
| reactions[reaction]['metabolites'] = eval(react_d[reaction][0]) |
There was a problem hiding this comment.
why using eval here? if there is a chance for an error, would not be better to have some catches for them ?
There was a problem hiding this comment.
This is because the metabolites are a dictionary that are saved as a string in the bigg_reactions.tsv database, arguably it would be better to save it as a dictionary in the file instead. For now I added a comment in the code (d7bf238).
dnngior/gapfill_class.py
Outdated
| Input: | ||
| list of reactions to remove from all reactions | ||
| Output: | ||
| None |
There was a problem hiding this comment.
That is a bit funny.
Consider having something like "trimmed list of candidate reactions "
There was a problem hiding this comment.
I updated the comment to read "list with reactions removed" (d7bf238) though technically the function does not output anything but modifies the list in place
I finally got around to fixing some of the last items of the features I added. As far as I can tell this should not break anything.
New Features:
It misses some of the later commits of the main branch (mainly related to what @hariszaf did last week I think). Hopefully we can fix this using GitHub. I also have not yet written anything about the CLI in the example notebook or readme as this will take some time. The medium and medium_file arguments might be redundant but I did not want to remove something.
Let me know if I missed something or do something differently