Fix hgvsg for insertion with ref allele -

[nakib103]

Problem

Here #1077 we have fixed hgvsg for multi-allelic variant. But it introduced a bug for insertion types.

We are not sending the end position during trimming -
https://github.com/nuno-agostinho/ensembl-variation/blob/3ed1013c5bfacb2416db9e67d50cb96fb2fcb0b6/modules/Bio/EnsEMBL/Variation/VariationFeature.pm#L1970

What it does for insertion is that it increase the end position by length of reference sequence -1 -
https://github.com/nuno-agostinho/ensembl-variation/blob/3ed1013c5bfacb2416db9e67d50cb96fb2fcb0b6/modules/Bio/EnsEMBL/Variation/Utils/Sequence.pm#L970

This becomes problematic if the length of reference sequence -1 is not the same as end - start and there is no trimming happening so that position sticks around. This happen only in the case when we have - in ref allele for insertion.

The end result is different position in slice and sub slice -
https://github.com/nuno-agostinho/ensembl-variation/blob/3ed1013c5bfacb2416db9e67d50cb96fb2fcb0b6/modules/Bio/EnsEMBL/Variation/VariationFeature.pm#L2016-L2019

and, terminating in a error here -
https://github.com/nuno-agostinho/ensembl-variation/blob/3ed1013c5bfacb2416db9e67d50cb96fb2fcb0b6/modules/Bio/EnsEMBL/Variation/Utils/Sequence.pm#L511

Fix

• Simply provide end position so trim_sequences do not try to calculate it.
• And, update the ref_end position. Otherwise it will fail in trim on reverse direction.

Test:

chr8 143694491 143694490 -/TT/TTT +
chr8 143694491 143694490 -/GG/GGG +
chr8 143694490 143694492 GTT/GT/T + (trim on reverse direction)

[olaaustine]

LGTM