Re: Science Aug. 2nd

Iddo Friedberg (
Fri, 23 Aug 1996 18:25:11 +0300 (GMT+0300)

This message is in MIME format. The first part should be readable text,
while the remaining parts are likely unreadable without MIME-aware tools.
Send mail to for more info.

Content-Type: TEXT/PLAIN; CHARSET=us-ascii
Content-ID: <>

Hi All,

Chao Tang has responded with answers to several questions that have arisen
during the discussion.



---------- Forwarded message ----------
Date: Thu, 22 Aug 1996 15:49:11 -0400
From: Chao Tang <>
To: Iddo Friedberg <>
Subject: Re: Your article, Science Aug. 2nd

Dear Mr. Friedberg:

Thank you for introducing me to the interesting discusions about our
in Science. In the following I (and my colleagues) tried to clarify
points raised in the discusion. I will continue to read the discusions
on the web site. Please keep it going. Let me know if I can be any
further help.

Best regards,


Chao Tang
NEC Research Institute Tel: (609)-951-2644
4 Independence Way Fax: (609)-951-2496
Princeton, NJ 08540 Email:


Iddo Friedberg ("\''/").___..--''"`-._ Phone: (972)-2-6585459/3 `9_ 9 ) `-. ( ).`-.__.`) email: (_Y_.)' ._ ) `._ `. ``-..-' web: _..`--'_..-_/ /--'_.' .' More info: finger Random quote: You're never too old to become younger. -- Mae West

--------------59E21CFB3F54 Content-Type: TEXT/PLAIN; CHARSET=us-ascii; NAME=sci-www Content-ID: <> Content-Description:

>"protein structures are selected because they are... stable against >mutations". What kind of mutations are we talking about here?

Random mutations. Proteins should be "relatively" stable against random mutations. On the other hand, we know that proteins are not stable against mutations on certain "conserved sites".

>Proteins are really only stable against _conservative_ mutations e.g. >polar->polar, hydrophobic->hydrophobic.

Depending on which site, mutations of hydrophobic <-> polar can be stable too.

>I think what follows from this is that on the average, a random point >mutation in a sequence leading to a highly-designable structure (HDS) is >less likely to unravel it than a random point mutation in a non-highly >designable structure. (NHDS). (Isn't this actually implicit in the >definition? More sequences encode an HDS than an NHDS?)

Right. But this is not necessarily implicit in the definition. A complete picture of N_s would include all possible mutations (e.g. multi-site correlated mutations), not just the random point mutation.

>Then again, HDS's have more conserved sites. (Fig. 4a). So I may be >missing something here...

No, NHDS's have more conserved sites, thus they are very unstable against random point mutations. Imagine a structure which can be designed by only one sequence, N_s=1. Then changing any of its sites (i.e. replacing an H with a P, or vice versa) would unravel the structure (the structure is no longer the ground state of the mutant).

> The authors observe that their "designable" structures are characterised > by having _both_ of the following kinds of sequences: > > 1. sequence families with many conserved positions (i.e. > conserved H residues, or conserved P residues) > > 2. some sequences completely unrelated to other sequences > (i.e. statistically insignificant sequence similarity - > e.g. no positions with conserved H across the family)

Well, from the Fig 4 of the paper one can see that there are some sites which are highly conserved while other sites mutable (H <-> P). I don't think that we have much evidence (if any) for the type 2 sequences in our study. I would guess that in larger (than 3x3x3 or 6x6) models, type 2 sequences would emerge.

>Being a physical chemist, the aspect that puzzled me was the coincidence >between stability to mutations(whatever that means exactly) and thermodinamic >stability >Is it just because of the entropy term ? >Or is it an artifact of the arbitrary choice of Epp, Ehh and Ehp

The thermodynamic stability we referred to is the energy gap between the two lowest energy (compact) states (ground state and the first excited state) for a given sequence. The gap for a structure (Fig. 3) is an average over N_s sequences which design that structure. There is no (configurational) entropy involved here. The gap is, however, related to the designability (N_s) of the structure. We gave an argument for this in the paper. The results are insensitive to the choice of parameters, as long as the choice satisfies some physical constraint (stated in the paper).

>Interesting paper with possible implications for many fields. For example, >if possibilities are limited by designability, does it mean that there >will be some structures and activities that will never be able to be made?

For structures, the answer is yes. For activities, I am not sure. This paper doesn't address this question.

>"It is the requirement that many sequences design a structure that >leads to formation of secondary and tertiary structure."

>It seems (to me at least) that this is highly unlikely to be a >correct statement. Why? Because I believe the reasons that secondary >and (some) tertiary structure form is already known (and it's not >at all related to the authors' hypothesis)... but I won't digress >further along this line...

What we mean is that certain structural motifs emerge from highly designable structures of this simple model. And these motifs are "protein-like". Of course, as pointed out in the paper, other factors as H-binding, geometry and sizes of amino acids, etc. will affect the detailed structure of proteins. But we believe that the designability principle should play a crucial role in the structures of real proteins.