[Biopython-dev] [Bug 2381] back-translate

[Howard Salis]

Yes, I know it's a one-to-many mapping. But that's why it's nice to
have a handy subroutine for doing it.

For codon optimization, all possible back translations with
unambiguous nucleotides would be best. Then, one evaluates some
objective function over all possible sequences to find an optimal one.
Optimality depends on the application, but eliminating restriction
sites, avoiding certain repetitive or transposon sequences, etc is
very common.

For searching for homologous proteins, it would be best to have the
back-translate function produce something that could be fed into an
alignment program or regexp expression. Then, one could align a
database of sequences with your back-translated protein to determine
which sequence is most similar to your protein. Basically, this is
what BlastP does (you might want to look up its algorithm to determine
a good way of doing this, if you wish to reproduce it in Biopython or,
otherwise, rely on the NCBI webserver).

What does the current back-translate function output?

-Howard

On 10/19/07, Peter <biopython-dev at maubp.freeserve.co.uk> wrote:
> Howard Salis wrote:
> > Yes. Back-translating a sequence is important in codon optimization,
> > searching for homologous proteins, etc.
>
> Unlike forward translation, transcription, back-transcription,
> complements and reverse complements, back-translation is not a
> one-to-one mapping.
>
> In your examples, would you want to know all:
> - all possible back translations (as unambigous nucleotides)
> - all possible back translations (as ambigous nucleotides)
> - a possible back translation (using ambiguous nucleotides)
> - a possible back translation (using un-ambiguous nucleotides)
>
> For example, back translating an Tyr => UAC or UAU => UAW (nice and
> clear - we can represent this perfectly with a single ambiguous codon).
> On the other hand, Arg => AGA, AGG, CGA, CGC, CGG, CGU => AGR or CGN
>
> Oh, and would you expect DNA or RNA back?
>
> Peter
>
>