2 replies [Last post]
Joined: 02/24/2011
Groups: None

Elevator pitch summary: reflections on intermediate structure and function in proteins.

I noticed an interesting shape in an intermediate stage of T-cell I made a slide show. Focused for the moment on how the proteins move rather that their precise structure.


phi16 and I have had a couple of chats about ideas like this. So in that spirit ...

The level at which we manipulate the proteins reminds me of the earliest days of computation - programming with a soldering iron and a plug board. That is hard coding.

Think of what we do as a search rather than a calculation. How many possible states of the protein are there? call it a bazillion. If you have to guess (a priori probability) we should find the solution after about a bazillion/2 steps. If we had an ordering system for the states that made the score a monotonic function of some ordering parameter then we could cut the search steps to something more like log(bazillion). a big gain. Probably can't make such an ordering in less than bazillion * log(bazillion) steps.

since bazillions are quite dauntingly big, the properly lazy man will seek an alternative, a heuristic strategy for choosing the states to look at.

Thinking about higher complexity component structures helps to reduce the complexity of the problem. Understanding a simple lego construction means understanding the way 10 blocks fit together rather than understanding all the connections between 500 items. How big a difference? the difference between 2^10 vs 2^500.

Certainly the characteristics of the lego block depend on the intimate details, the physics of the block. But the block is not merely a natural object. It is also an artifact. The physics of a lego block are what they are because someone/thing made a lego block. The maker put the components together and formed them so that they could be lego blocks.

Proteins in cells are made. They are artifacts. Understanding how and why a thing is made gives rise to a set of ideas and ways of classifying and grouping complex parts, of coming to see them as one thing - a chain perhaps, or a wall, or a hollow ball. Thinking about such components tidies the complexity which sounds neat and dainty but is actually ferocious. Each reduction of complexity by m has the effect of dividing the bazillion by 2^m. Leverage.

To scale the relative importance in an unimaginative way - n searchers cuts the time to bazillion/2*n. The real value of human hunters is not so much numbers directly it is numbers of insights. Insights, ways of eliminating complexity, will rule.

My plan? Muhahahaha! Reduce complexity to get to log(bazillion).

I have also begun to write a bit about foldit on my wiki - wut. you are welcome to visit.
please let me know if you have issue with anything I touch on there.


I appreciate comments suggestions and yes even corrections 8P.

Joined: 02/24/2011
Groups: None
Spmm's post

Spmm posted http://fold.it/portal/node/989422 .

I think it is very extremely interesting. on whall street they call this a table thumping pitch.
follow the links.
look at what these big proteins do - why they are how they are.
our focus- scores etc are based on a static pictures of how things are at an instant.
but really the things we are looking at are in that static state because it is a frame in a well defined quantum dance from one state to another.

as quantum number gets large the world gets classical and continuous. when n is small, the world jumps and jerks and is very finite. We work in the middle - more than a few but less than infinite. At the right scale

there is a threshold scale where quanta become just stable enough to become predictable. This is the result of aggregation. Aggregation reduces the total possible micro-states. It makes some

our ui, and our vision of the molecules is static. Euclidean where. S3. The proteins are artifacts built for S3 x t.

Joined: 02/24/2011
Groups: None
sorry the above got a bit

the above got a bit garbled as I was interrupted and needed to save what I'd written without completing it.
wish I could edit the post.


Developed by: UW Center for Game Science, UW Institute for Protein Design, Northeastern University, Vanderbilt University Meiler Lab, UC Davis
Supported by: DARPA, NSF, NIH, HHMI, Amazon, Microsoft, Adobe, Boehringer Ingelheim, RosettaCommons