[cowboysauce]

in theory it's possible, in practice it's not. DNA encodes the information on how to produce to proteins. Proteins can catalyze chemical reactions. But there's no straightforward to look at a nucleotide sequence and figure out what it's corresponding protein does. Nor can we design proteins with arbitrary functions.

The entire thing is so unimaginably complex. For example, for a lot proteins that are catalysts (aka enzymes) the actual catalytic part is a metal ion and the protein mostly provides scaffolding. A nucleotide sequence alone doesn't directly tell you what ion is needed. In some cases, multiple ions can fit, but only one actually results in the protein work. This the basis for how a lot of toxic metal exert their toxic effects.

It's also not as simple as a nucleotide sequence codes for a protein and that's it. Proteins fold into their final shape from the chain of amino acids that DNA encodes. Protein folding in general is a hard problem. Biological proteins may have other proteins (called chperonins) that help them fold into their proper configuration. Then proteins may also be modified after they've folded (again by other proteins). Some proteins are made up of multiple sub-units as well.

[xg15]

Thanks a lot for the info.

> Proteins fold into their final shape from the chain of amino acids that DNA encodes. Protein folding in general is a hard problem.

My hope was that we had made progress in exactly that domain. Yes, the relationship between a nucleotide sequence and the resulting protein is extremely complex, but my impression was that we have tools to simulate the folding process for a given chain of amino acids (Folding@Home comes to mind).

So I was imagining a brute-force like process, where you (somehow) start with some candidate sequences, simulate how they would fold and use the sequence that comes closest to the molecule you wanted to have in the first place. Of course this only works if your target molecule can be assembled out of amino acids.