what database should I use for de novo genome in AUGUSTUS

Question

[Jl]

I'm trying to annotate lizard genome and want to use AUGUSTUS to find protein sequences. Can I used the chicken database in AUGUSTUS? How can I make the result be more accurate? Thank you!

Answer 1

[lieven.sterck]

You can if you want to (technically you can likely go for any one of them) yes

however as you already suspect the results might not be super accurate. To get them more accurate you to train Augustus on your specific genome (is there no auto procedure for that nowadays, using the busco genes or such for instance?)

If that is too much, you should take the available database/trained-model that is most similar to your genome of interest. By that I don't mean from an evolutionary point of view but rather from a genome composition point of view. So more a genome with roughly similar gene structures (think of #exons/introns, length, ... ) , similar %GC , characteristics like that.

for providing protein match info this is less crucial, you can use any that makes sense or even a combination of databases. (the more the better is the advice here as long as it makes sense)