Csv convert for greentable format

[djarecka]

I've modified the converter from #19 to work for the examples of the Green Table (for now I only updated cvs2yaml.

I've added also an example of cvs files and the header that could be used to run it: python utils/csv2yaml.py --csv csv_test/GreenTable.csv --enum csv_test/GreenTable_ValueSet.csv --header csv_test/head.yaml --output csv_test/gt_test.yaml

The output should look like csv_test/GreenTable_model.yaml

I've slightly modified the GreenTable format, see here (note that I only worked on Generate Library for now). The changes are related to the Value Sets and I've created a new tab: Generate Library Value Sets (you can compare with Value Sets Orig). The original tab for Value Sets was hard to automatically convert to pandas.DF, so I've added column Field Name to remove the titles of previous multiple columns. In order to properly link ValueSets from the Generate Library to the values for specific Field Name, I assumed that it is either the original Field Name from the Generate Library tab or DataType has the name of the set values, e.g. ValueSet: barcode name.

Issues with Value Set that need feedback:

• [ ] I didn't know what to do with the Code and Code general columns, should this be added to the model?
• [ ] There was sometimes additional info in a column without a name, so I created Other Info, not sure what to do with it
• [ ] I didn't know how to interpret columns from R1/R2 index name, so for now I removed it

Other issues to discuss:

• [ ] I only added the fields that had "metadata" in the Identifier/Metadata column. I assumed that the additional parent class will have the name
• [ ] Some fields had integer as a type, but the example contained floats (when you run the code, you will see errors in the standard output)
• [ ] the examples from Enriched cell source barcode name do not match the value sets
• [ ] tried best to automatically split the examples, but probably we should have a better standard way of providing it

Other TODOs (that do not require feedback):

• [ ] add a way to update an existing classes (now I only read a header)

FYI. I will look into the code from the Yale team, I only opened the output quickly and I noticed that there are some things that we discussed done differently, so I decided to finish this PR to help with the discussion.

[djarecka]

@satra @lydiang @patrick-lloyd-ray - if you have time this week you can give me some feedbacks, especially on the issues listed in the PR description. If not we will discuss it later

[djarecka]

notes from the discussion with Pam at the SFN:

• R1/R2 index: the info in the "index-" columns shouldn't probably be included, but perhaps the specific values could have meaningful descriptions
• cell_source_barcode_name: we have to check is this really should be a set_value, there is still issue that the example doesn't match
• some of the value_set list should come from specific documentation (e.g. 10X), it should be documented and can evolve
• should be compared with the doc from them metadata repo

[patrick-lloyd-ray]

@djarecka could you check the integer data types here: https://github.com/brain-bican/metadata-schemas/pull/31 and make sure we are accurate?

[djarecka]

@patrick-lloyd-ray - yes I'm planning to check it, I forgot that this is in PR and that's probably better place to add more comments

[patrick-lloyd-ray]

@patrick-lloyd-ray - yes I'm planning to check it, I forgot that this is in PR and that's probably better place to add more comments

Great! Feel free to simply suggest changes directly on the PR.