[RC5] Protein folding. A new project???

Alan Gaul adgaul at netins.net
Thu Jan 25 21:09:23 EST 2001

I have a relative who is VERY heavily involved in topics related to protein
structures, with extensive experience using NMR (nuclear magnetic resonance)
equipment to determine the structure of similar molecules.  I relayed the
question about protein folding over the Internet, and thought the D.NET
group might be interested in the response:

"I just read over the protein folding site.  The problem of protein folding
is probably one of the most complicated problems in biochemistry right now.
Pretty much all of the computational methods used to study this are
approximations of just about everything because the computing power is never
enough.  (The project I did..[PHD thesis]..was just a small peptide, with
constraints which helps immensely, and then modeled into a protein.  The
peptide probably took me 12 hr. runs X 2 for 200 rough structures, and then
12 hr. X 20 for refining, and then much more to do the docked structure.
And usually when you try to do dynamics, motional movements, in a protein,
you set a small sphere of allowed motion and hold the rest constant.  Trying
to handle a whole protein is too computationally intensive.)

They are trying to handle this more rigorously, which is ambitious with
today's computers, but they are receiving a fair amount of interest since
they are publishing in places like Nature-Structural
Biology.....Scientifically, it is probably a very worthwhile investment.

In terms of why does anybody care?  I would definitely put it above
calculating prime numbers and chess games.  A protein is just a sequence of
amino acids and without doing NMR or crystallography, there is no way to
know what it looks like in a 3-D fold.  Predicting how a protein will fold
based on the sequence has been a problem for a very long time and probably
will be for a long time.  (They have contests every year where groups use
their computer programs to predict a protein structure while some other
group solves it experimentally.)  Protein structures teach us a lot about
how a protein works and why.

The thing everybody gets most excited about is simply structure-based drug
design (or I do anyway).  For example, the protein I worked on is an
oncogene which means one specific mutation and the protein causes lymphatic
cancer.  If you know how the protein (an enzyme) is folded, you can use that
structure to model in potential drugs.  Places like Abbott Labs are
using structure-based drug design to a high degree of success.

The whole human genome project is also now giving way to what's called
structural genomics.  That means, now we have a bunch of gene sequences that
we can translate into protein amino acid sequences, but we have no idea what
any of the structures look like.  Structural genomics is mostly brute force
trying to solve thousands of structures by crystallography and/or NMR (a
very daunting somewhat controversial task).  If someone could accurately and
reliably predict the structures computationally, life would definitely be


I usually just "lurk" on the sidelines here, and cannot claim any real level
of expertise on distributed computing.  I've always been impressed by the
efforts put forth by the D.NET group, and it certainly appears that D.NET
has a significantly more mature system for managing large distributed
projects.  Any new project obviously represents a huge effort for a
volunteer organization, so the goals should be selected carefully.

The Stanford group appears to be very knowledgable regarding protein
folding, and has source code posted for at least part of their algorithm.
They are building a distributed network to work on a single project.  In
contrast, D.NET has significantly more expertise in distributed computing,
with a system in place proven to handle multiple large projects
simultaneously and experienced programmers capable of optimizing code for
this system.  I think it would be a good match if a way could be found to
assist in their efforts without losing our identity as a group.

My thought would be for D.NET to work with the Stanford group to create a
client specific to our existing distributed network.  The workload would be
distributed the same way the RC5 and ORG projects are handled now, with the
results from our many volunteers tallied on the stats box the same way.  The
twist would be that the final results would be submitted to the Stanford
project as a single large contributor in D.NET's name.  We would essentially
be a VERY large team, and would probably rank #1 on their statistics very
quickly.  Given the press their project is likely to get, this would create
additional publicity for D.NET that might help draw more volunteers
interested in our other projects as well.

Just my $0.02 worth.....

> -----Original Message-----
> From: owner-rc5 at lists.distributed.net
> [mailto:owner-rc5 at lists.distributed.net]On Behalf Of Sanford Olson
> Sent: Sunday, January 21, 2001 8:46 PM
> To: rc5 at lists.distributed.net
> Subject: [RC5] Protein folding. A new project???
> This group at Stanford seems to be building a distributed
> computer network
> from scratch.  Also, their client software is somewhat buggy.  Perhaps
> Distributed.Net should help them out??  Seems like a very
> worthwhile project.
>  >>PROJECT GOALS: Solving the protein folding problem
>  >>
>  >>Understanding how proteins self-assemble ("protein folding") is a holy
> grail >>of modern molecular biophysics. What makes it such a great
> challenge is its >>complexity, which renders simulations of folding
> extremely computationally >>demanding and difficult to understand.
> Check out this web site:
> http://www.stanford.edu/group/pandegroup/Cosm/index.html
> --
> To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
> rc5-digest subscribers replace rc5 with rc5-digest

To unsubscribe, send 'unsubscribe rc5' to majordomo at lists.distributed.net
rc5-digest subscribers replace rc5 with rc5-digest

More information about the rc5 mailing list