Standardise SV types across inputs: VEP, VCF and Region

[nuno-agostinho]

ENSVAR-5226

Changelog

• Validation of each format is now encapsulated in the respective file
• All SO terms are now standardised and available in Parser.pm for any input: INS, DEL, TDUP, DUP, CNV, INV, BND (including if encapsulated in <>)
  • VEP_input now accepts the same terms as VCF, including INV, BND, <CN3>, <DEL>, etc.
  • Given the validation restrictions, Region accepts most terms, except if encapsulated in <> like <CN0> and <INV>

Discussion

Does it make sense to accept all SV types in all types of inputs? For instance:

• What does INS do if we don't have the inserted sequence in VEP_input/Region?
• What is the purpose of CNV without specifying number of copies?
• BND can be specified without mates, but VEP currently does not support breakend variants (not for long if we merge Add support for breakend variants from VCF #1399)

Testing

• I added unit tests to test new changes, so please check if VEP works with some random example inputs
• VCF input should return the same result before and after the changes
• Test that VEP_input accepts the new terms. Example:

1   20000     30000     <CN4>   +    cn4
1   881907    881906    -/C     +
5   140532    140532    T/C     +
12  1017956   1017956   T/A     +
12  1017956   1017956   INV     +    inv1
2   946507    946507    G/C     +
14  19584687  19584687  C/T     -
19  66520     66520     G/A     +    var1
8   150029    150029    A/T     +    var2

• Test that Region accepts the new terms. Example:

chr21:10-10:1/A
chr21:10-20:1/DUP
21:25587759-25587769/INV

[dglemos]

Merged into release/110 and main