Update Jannovar to 0.41 to fix incorrect MT annotations

[julesjacobsen]

The mitochondrial codon table is slightly different to the standard eukaryotic table:

2. The Vertebrate Mitochondrial Code (transl_table=2)

    AAs  = FFLLSSSSYY**CCWWLLLLPPPPHHQQRRRRIIMMTTTTNNKKSS**VVVVAAAADDEEGGGG
  Starts = ----------**--------------------MMMM----------**---M------------
  Base1  = TTTTTTTTTTTTTTTTCCCCCCCCCCCCCCCCAAAAAAAAAAAAAAAAGGGGGGGGGGGGGGGG
  Base2  = TTTTCCCCAAAAGGGGTTTTCCCCAAAAGGGGTTTTCCCCAAAAGGGGTTTTCCCCAAAAGGGG
  Base3  = TCAGTCAGTCAGTCAGTCAGTCAGTCAGTCAGTCAGTCAGTCAGTCAGTCAGTCAGTCAGTCAG

Differences from the Standard Code:

        Code 2          Standard

 AGA    Ter  *          Arg  R
 AGG    Ter  *          Arg  R
 AUA    Met  M          Ile  I
 UGA    Trp  W          Ter  *

The Jannovar version used in Exomiser versions <=13.2.1 only uses the standard codon table. This means that in some cases a mitochondrial variant can be incorrectly annotated, e.g. NC_012920.1:m.15150G>A is p.(Trp135*) using the mitochondrial codon table but p.(Ter135=) with the standard table. This leads to the change being classified as a SYNONYMOUS_VARIANT rather than a STOP_GAIN.

Solution

To fix this issue update to the latest Jannovar version 0.41

Side-effects

The update includes other changes to the Jannovar data model to handle gaps in the refseq alignments, which makes the existing Jannovar data unreadable by a newer Jannovar version. Exomiser uses a protobuf representation of this data which should mitigate this but there needs to be extensive testing about wethere older Exomiser versions can safely read newer data and the updated version can read old data.

[julesjacobsen]

Fixed in 24ee98b4a7cffa9e394a3c145853c34f505b5310