Changing genetic code table

[KatharinaHoff]

It is currently not possible to switch genetic code table for running BRAKER. AUGUSTUS can use a different genetic code, but GeneMark-ES/ET must be extended before BRAKER can be extended.

BRAKER code extension is planned, depending on GeneMark-ES/ET extension.

Required changes:

• command line option for genetic code table in BRAKER

• translation script prior aa2nonred.pl (part of AUGUSTUS)

• editing AUGUSTUS parameters prior training and prediction

• command line option for getAnnoFastaFromJoingenes.pl in BRAKER

[KatharinaHoff]

Linking howto change translation table for AUGUSTUS: https://github.com/Gaius-Augustus/Augustus/issues/26

[OnlineArts]

That's a great idea!

Recently, I figured out that the organism of my genome uses TGA for Trypthan (W). GeneMark-ET just damage all prediction by assuming that's a stop-codon.

I kicked out all TGA stop codon related perl code in GeneMark-ET. Finally, there is still a link in the gmhmme3 binary file. I fear this really helpful enhancement for BRAKER will not exist soon.

Further, only changing Augustus has a minor effect on my data but it's still helpful. I suggest to add an option for the /CONSTANT/ parameters as you already mentioned in Gaius-Augustus/Augustus#26. Additionally, I suggest to add translation_table as suggested by Augustus (chapter 4).

Currently, I'm trying to "overwrite" all TGAs to TGGs in the genome and will compare the final results. Maybe I could regain information by "improving" the original prediction.

[KatharinaHoff]

With commits https://github.com/Gaius-Augustus/BRAKER/commit/1246074a41b1547dbae2374d320ec3edd5f850ea and https://github.com/Gaius-Augustus/BRAKER/commit/11581dce27383e2f46c61c9b56f9e9ebd374c52b , the following features were added to braker.pl:

• command line option for genetic code table in BRAKER (still under development options): --translation_table=INT
• translation prior aa2nonred.pl in helpMod.pm was modified; some codons are not ideal because they can either be translated to a stop or to an amino acid, in these cases, we always translate to stop; this is not the most crucial part though because the generated protein sequences are only used for the redundancy filter prior training AUGUSTUS
• setting stop codon probabilities in AUGUSTUS parameters prior training and prediction according to translation table
• command line for getAnnoFastaFromJoingenes.pl in BRAKER now contains translation table

BRAKER has been prepared for the translation tables 11, 6, 10, 12, 25, 26, 27, 28, 29, 30, and 31. These are all the nuclear tables that I am aware of.

@OnlineArts Your suggestions are implemented. Thx.

@tomasbruna : whenever GeneMark-EX contains an option to alter the translation table, please let me know.