ncbi_egapx: form seems to work and generates what looks like valid yaml

[fubar2]

needs tests but cannot even run let alone test here - no machine with 120GB or 31 cores - because of the resource requirements baked into the docker image

[richard-burhans]

Looks good!

[richard-burhans]

Attention: deployment failure!

https://github.com/richard-burhans/galaxytools/actions/runs/10781909881

[nekrut]

@fubar2 or @richard-burhans = we need to add an option that would allow EGAPx to take a file with Protein FASTA as annotation source (see https://github.com/ncbi/egapx?tab=readme-ov-file#input-data-format)

[fubar2]

would allow EGAPx to take a file with Protein FASTA

@nekrut: If that protein fasta is independent of the NCBI, then it may make sense to use it in the HMM.
Easy to add another form input, but you probably need to be certain that the supplied protein annotation is statistically independent of the NCBI protein annotation to contribute useful, unbiased information.

If any NCBI protein fasta exists for a taxon, it is AFAIK an output from running the internal NCBI pipeline that has become egapx. So predicting proteins using egapx, relying on information from a fasta that has been predicted by the father of egapx, may yield biased and uninterpretable results AFAIK because of lack of statistical independence in some of the inputs used for prediction?

Not an expert but this is a good question for one of the NCBI authors.

[marco91sol]

I believe we can implement this and run some tests to compare the results with and without the additional protein FASTA and HMM files. EGAPx treats these as optional parameters:

However, the challenge is that the taxid-protein set is not available for all taxa (I've already contacted the CGR group for more information). For instance, I encountered this issue while testing on sharks, where the taxid wasn't connected to any protein set.