Re: poly: Nanomedicine and cell types

In a message dated 5/29/00 8:12:10 PM Pacific Daylight Time, hal@finney.org
writes:

>Reading ahead to section 8.5.2.2, I came upon a puzzling claim.
>"Complexity theory and phylogenetic comparisons suggest that the maximum
>number of cell types Ncell ~ Ngene^(1/2) = 370 cell types for humans
>with Ngene ~ 10^5 genes."

> The statement is footnoted with two references: Stuart Kauffman's book
> "The Origins of Order", and a technical paper by Kauffman.
>
> Is anyone familiar with the logic which leads to the conclusion that
> the number of cell types would be roughly the square root of the number
> of genes?

The "complexity thoery" is just some computer simulations.
Kaufmann showed that in certain types of randomnly connected
networks, the number of alternative stable cycles is
approximately the square root of the number of elements.

The "phylogenetic comparisons" is a regression of # of cell
types vs. DNA content, which yields roughly a square root
function. Note: DNA content, *not* gene number (as the
theory would suggest). It doesn't work for gene number.

My own attempts to noodle around with this came up with
different conclusion than Kauffmann's. I found that more
heavily interconnected networks never settled down to
short-cycle attractors, while less heavily interconnected
networks would basically only cough up one cycle. Other
cycles were there but very unlikely and not very
accessible. My tentative conclusion was that *purely*
random networks are a bad model for differentiated cell
types.

Kaufmann has been pretty influential in biology but this
particular idea has engendered virtually no work. I've
looked and just not been able to find any. So developmental
biologists aren't too interested in this idea, whatever
that interest is worth. I suspect the reason is that
differentiation doesn't look at all like the interactions
of complex systems with basin shifts - it's more of a
series of binary decisions, each mediated by a small set
of specific genes that play little role once the decision
is made.

This might not be entirely fair to Kauffmann's idea as the
*maintenance* of cell types does look more like interaction
of complex sets (of promoter/enhancer elements). I haven't
been able to find anything on that either. I've tried to
discuss it with the professors I had for transcriptional
regulation. With one I couldn't get the idea across;
the other thought it wasn't interesting.
Received on Mon May 29 22:25:42 2000