3

Are there any algorithms already in existence for reverse protein folding/docking (e.g. taking a desired surface map/docking profile and generating DNA which will produce a protein to fold into it?)

Protein folding seems straightforward (though computationally intensive) and protein docking is likewise straightforward, but thus far I haven't come across any protein unfolding algorithms, though I might just be searching for the wrong keywords.

Any advice in this area?

llrs
  • 4,693
  • 1
  • 18
  • 42
CoryG
  • 195
  • 1
  • 3
  • 1
    This is the holy grial of bioinformatics! Predict the structure of a protein from the amino acid chain is hard and has many errors usually, but doing the other way round is almost impossible AFAIK. But if you found something let me know :) – llrs Nov 22 '17 at 21:43
  • @Llopis just starting on it to be honest. I've been wanting to make a useably abstracted programming language which compiles into actual organisms for a long time and always looked at it as something best approached by cross referencing semantic-style matching algorithms vs DNA with tagged trait data by species (and where possible by gene,) but I recently decided to go get a biotech degree and it elucidated much of what I was missing - it seems protein unfolding is the only step actually missing to employ standard programming techniques to get it done (plus we already have DNA printers.) – CoryG Nov 23 '17 at 09:37
  • 1
    Read this article and the articles that reference it. It is an utopia to program and compile organisms that way. As one of my professors said "the paper holds everything but then you need to check it". Go step by step if you want to compile organisms. – llrs Nov 23 '17 at 09:46
  • Oh I'm pretty sure I have the details at this point, the issue is the protein unfolding - everything beyond that isn't much different from logic used in systems programming. I'll check out the paper though, thank you. – CoryG Nov 23 '17 at 09:59

1 Answers1

1

The researchers behind FoldIt have the occasional design challenges that use human game players to help design proteins to fit a structure.

However, even with a known protein sequence there's still a lot of labwork involved in trying to find a good working DNA sequence. Just generating the DNA sequence is a difficult task due to redundancy in the genetic code (i.e. 64 codons for 20 amino acids + stop signals). When that is combined with cell-specific tRNA frequencies that can influence the timing of amino acid incorporation (and therefore how a protein folds as it is created), creating a sequence that reliably fits a shape ends up being quite difficult.

gringer
  • 14,012
  • 5
  • 23
  • 79