arq5x / gemini

a lightweight db framework for exploring genetic variation.
http://gemini.readthedocs.org
MIT License
318 stars 120 forks source link

removal of little-used tools #787

Open brentp opened 8 years ago

brentp commented 8 years ago

updates such as my recent effort to move everything to python3 take an amount of time that is super-linear with the amount of code. We have a lot of un-used/little-used tools. I propose that we remove them and make it easier for users to create their own modules as needed to reduce the burden on gemini while improving flexibility.

Here are the current tools: {actionable_mutations,amend,annotate,autosomal_dominant,autosomal_recessive,bcolz_index,browser,burden,comp_hets,db_info,de_novo,dump,examples,fusions,gene_wise,interactions,load,load_chunk,lof_interactions,lof_sieve,mendel_errors,merge_chunks,pathways,qc,query,region,roh,set_somatic,stats,update,windower,x_linked_de_novo,x_linked_dominant,x_linked_recessive

I propose that we remove (in order of my preference to remove).

this would leave: {amend,annotate,autosomal_dominant,autosomal_recessive,browser,comp_hets,db_info,de_novo,dump,gene_wise,load,load_chunk,mendel_errors,merge_chunks,pathways,query,roh,update, x_linked_de_novo,x_linked_dominant,x_linked_recessive}

roryk commented 8 years ago

I am down for removing anything you need to make your life easier. AFAIK, the only one of those we use anywhere near non-rarely is --stats.

brentp commented 8 years ago

Hi Rory, thanks for the feedback. is there any particular tool within stats that you use?

roryk commented 8 years ago

Just the simple transversion stats and the genetic distance, something we could easily make a little module for.

arq5x commented 8 years ago

The region tool gets some use and the set_somatic, actionable_mutations, and fusions tools were part of the speedseq paper.

I agree with deprecating these, but perhaps we should start a gemini utilities repo to house the ancillary tools/modules?

jxchong commented 7 years ago

We definitely use region and sometimes stats but stats can be easily replicated with other existing tools.