Back translate

[jfjlaros]

Added back translate interface.

[jfjlaros]

Note that I use the NM to NP lookup table form the db module, this table is not symmetric so perhaps I am using it wrong.

[martijnvermaat]

I rebased and squashed this branch on current master, makes reviewing easier.

[martijnvermaat]

I'm wondering a bit about naming, i.e.,:

1. Backtranslate versus back-translate versus back translate.
2. Translate versus translator versus translation.

And in code:

1. backtranslate.BackTranslate suggests one word and two words.
2. mutalyzer.back_translate takes another approach.

[martijnvermaat]

I don't understand this. Is this a TODO before we can merge this?

[jfjlaros]

No, I don't think this is blocking.

1. Currently the back translator only supports NM or NP. In principle other references (NG, NC, LGC, ENSx) can also work.
2. This refers to the name checker. We currently do not check whether the reference amino acid is actually there. E.g., for p.Glu1Arg we do not check whether there is a glutamine at position 1.

[martijnvermaat]

@jfjlaros I revised this PR and fixed most of the issues I could find.

I just commented with some small remaining questions. Before merging this I would also like to include some tests.

[martijnvermaat]

From an NP finding the corresponding NM is currently sort of broken when the NM is already in the database. This is because the database stores the NM including its version number and we query for the NM without version number.

It still sort of works, since the NCBI will now be queried, but of course we would like to use the cached NM.

I think we should first fix the database to store the versions separately.

Note that this is only picked up by the unit tests when disabling the internet connection.

[martijnvermaat]

After some tests it looks like we can actually have transcript-protein links with version numbers, so we should probably do that first.