My hypothesis is the following. There is a transformation from the made up chemlambda molecules, which TRANSFORMS:
– node into real molecule
– port into real molecule
– bond into real molecule
and some other real molecules called here “enzymes”, one per any type of graph rewrite
– graph rewrite G which replaces this configuration LT of two nodes and 1 bond into that RT configuration TRANSFORMS into the chemical reaction between enzyme G and the transformation of LT into real chemicals, which gives the transformation of RT into real chemicals and the enzyme G (perhaps carrying away some other reaction products, to have conservation of # atoms)
The argument for that hypothesis is that the rewrites are so simple, compared with real chemistry of biomolecules, that there have to exist such reactions.
This is explained in the Molecular computers
Suppose that the hypothesis is confirmed. Either by identifying the TRANSFORM from scratch (i.e. by using chemistry knowledge to identify classes of reactions and chemicals which can model chemlambda), or by finding the enzymes G and the real molecules corresponding to node, port and bond in some fundamental biochemical processes (that would be even more wonderful).
Suppose this. What then?
Then, say I have the goal to design a molecule which does something inside a cell, when injected in the body. It does it by itself, in the cell medium. What it does can always (or in principle) be expressed as an algorithm, as a computation.
I use chemlambda and TRANSFORM to design the molecule and check that once I have it, it does the job. It is of course a problem to build it in reality, but for this I have the printer of Craig Venter, the digital biological converter https://youtu.be/DVL1nL3SU6I .
So I print it and that’s all. When injected in the body, once arrived in the cell, it does the job.
Other possibilities would open in the case some formalism like chemlambda (i.e. using individual molecules and rewrites, along with trivial random algorithms) is identified in real biochemistry. This would help enormously the understanding of biochemical processes, because instead of working empirically, like now when we work at the level of functions of molecules (knowing well that the same molecule does different things in different contexts and that molecule-function association is very fragile in biology), we might work inversely, from using functions as black boxes to being able to build functions. Even to go outside functions and understand chemistry as computation directly, not only as a random medium for encoding our theoretical notions of computation.
See more about this at the chemlambda index