Add options to trident init/rectify CLI

[TCLamnidis]

Currently, trident init can be used to set POSEIDON.yml attributes like snpSet, but cannot be used to set the referenceGenomeAssembly or referenceGenomeAssemblyURL attributes of the dataset, even though that is something that is closely associated with the genotype data added in a package as well.

Similarly, trident rectify can be used to update other attributes of the YAML, like contributors, or checksums.

Currently, referenceGenomeAssembly* can only be updated by manually editing the YAML file.

I would favour adding options like --referenceAssembly and --referenceGenomeAssemblyURL to trident init, so this information of the genotype data can be added to packages directly on the CLI. Similarly, it would be nice to have an option to rectify existing packages to add/edit this information (if the package is of a compatible version) from the CLI.

I generally find that having to open and edit the POSEIDON.yml files manually is error prone.

Unsure how simple that is to implement, but I could also imagine a system like that used for contributors with a mini-language using markdown link syntax to set the assembly name and URL also.

[nevrome]

Makes sense to me - and at least adding these options is trivial. I guess this would be a good opportunity to finally implement #343.

What about the following order of tasks:

1. Merge Add xerxes codebase #369 and add a fix for trident genoconvert removes referenceAssembly information from POSEIDON.yml #371.
2. Implement Rectify / Modify change #343 with referenceGenomeAssembly options for modify.

What do you think, @TCLamnidis? The second task is not too difficult to implement. This could be your first Haskell project 🙂

[stschiff]

Yes, OK.

First, fully agree with @nevrome that we should get Xerxes into the codebase first.
Second, I'm happy to implement whatever @TCLamnidis needs as Minotaur is our heaviest user! Having said that: While I agree that manually editing YAML files is not good, there would be a better way: You could just load the YAML contents into python/R as structured data, edit at will in-memory (adding fields and such to record-types, no problem), and then writing that out again as YAML. In python this would be something like 6 lines of code, and you would then not anymore be dependent on us implementing every single field as a CLI option to trident. You could even just completely ditch the automatically created POSEIDON.yaml by init and simply create it fully yourself. That's why we have a Schema, to make these things doable outside of our tools. Just my 5 cents... but again, happy to just implement as many of these on the CLI as you prefer.