Merge functional annotation file on HGVS, not original_mutation_description

[miseminger]

The guidance for the DataHarmonizer template's 'original_mutation_description' field says any string can be inputted.

At the moment, addfunctions2gvf.py relies on the contents of 'original_mutation_description' matching names derived from the VCF, which could result in data loss:

merged_df = pd.merge(gvf, df, on=['original_mutation_description', 'protein_symbol'], how='left') #, 'alias'

[miseminger]

See also:

nf-ncov-voc/bin/functions.py

Lines 439 to 444 in 49ac6c1

	# create a "Names" column that holds the amino acid name (minus 'p.')
	# if there is one, or the nucleotide level name if not
	df["Names"] = df["hgvs_nucleotide"]
	protein_mask = df["hgvs_protein"].str.contains("p.")
	df.loc[protein_mask, "Names"] = df["hgvs_protein"].str.replace(
	"p.", "", regex=False)

[miseminger]

Solution: merge on HGVS, and don't add the original_mutation_description to the GVF from the annotation resource at all. This brings up one last issue: The functional annotation resource and the GVF can't both contain a key called 'original_mutation_description' that mean different things. Proposed: change 'original_mutation_description' in the VCF to 'name'?

[miseminger]

Functional annotation spec: https://docs.google.com/spreadsheets/d/16RBFfz-GTK6cN5rxQA-Z9H-OKEcRqTkOWuJb5AOz1Eg/edit?gid=186254504#gid=186254504

[miseminger]

In the functional annotation file, we have:

column name	example value
nucleotide mutation	NC_045512.2(GU280_gp01):g.1438T>C
amino acid mutation	YP_009724389.1:p.Phe480Leu

In the (updated SnpEFF) GVF, we have:

column name	example value
hgvs_nt	NC_007362.1(FLUAVH5N1_s4gp1):g.1338T>C
hgvs_aa	YP_308669.1:p.Val446Val

If we transform the SnpEFF HGVS back to SnpEFF's former version like we decided for #211, we won't be able to merge on the nucleotide HGVS names--which is mostly fine, we can merge on the amino acid names, except for intergenic regions. Also, though, it's unlikely that functional annotations will be for intergenic regions very often.

@anwarMZ

[miseminger]

Solution: match on hgvs_aa and/or hgvs_alias (only one match required for merge).