Perl Conference Handout

Steven E. Brenner brenner@hyper.stanford.edu
Fri, 8 Aug 1997 11:59:53 -0700 (PDT)


Steve,

  Good thoughts.  I incorporated your comments for the handout.


> 
> >From our discussion of 2D structures in July, I think there is agreement  
> that the Struct module should not deal with both 2D and 3D structures. 
> Objects that represent a molecule's 2D structure would be handled by a 
> different module which could be associated with either a 1D or 3D 
> structure by a delegation mechanism.
> 
> Some more thoughts on the 2D vs. 3D issue: 
> 
> 3D structures *contain* 2D information, so a 2D structure object should 
> be derivable from a 3D structure object. If you believe Anfinsen, 1D 
> sequences contain 3D (and therefore 2D) structural information. The point 
> here is that there is redundancy in biological data. To be really useful, 
> the Bio:: modules should account for this, keeping the violations of good 
> software engineering principles to a minimum. Although tempting, we 
> should not sacrifice biological robustness for the sake of good s.e.

  However, while 1D defines 3D, we don't know how.

  I guess the question is does a sequence "hava a" structure or does a 3D
structure "have a" sequence.   I guess, really it is a molecule which "has
a" sequence, and (2d and 3d) structure.  

  Basically, I think that the objects should model biology as practically
used, rather than biology as it presumably exists.  Therefore, knowing 1D
sequence does not imply knowledge of 3D structure (even though
biologically, it does).


Steve