Protein similarity via hydropathy plots?

Bernard Murray bernard at elsie.nci.nih.gov
Thu Apr 7 22:18:25 EST 1994


In article <940407155254.1a68 at jason.uthct.edu>, SHAUN at JASON.UTHCT.EDU ("Shaun D. Black") writes:
> Michael Coyne asked how to discern if two proteins are related by looking 
> at their hydropathy plots.  Yes, they looks like a bunch of wiggles, but 
> there's gold in them thar hills.  In fact, Sweet and Eisenberg (J. Mol.
> Biol. 171: 479-488 (1983)) showed that correlation of sequence 
> hydrophobicity measures similarity in the 3D structures of proteins.
> So, even though we have no robust way to interpret the 'wiggles', if two 
> proteins share similar patterns, it makes a powerful statement about the 
> folding potential being similar.

...Details on methodology deleted for brevity.

Why not simply align the sequences in which you are interested?  By making
these plots you are simply parameterising the amino acids and then using the
values to generate something for easy (visual) comparison.  In doing this you
bias the analysis by choosing that scale.  Is there a rational basis for
picking hydrophobicity, as opposed to any of the other hundreds of scales
available (eg. see Kidera et al. - J.Prot.Chem. 4, 23-55 [1985] for a
comparison of a mere 198 parameters)?  Even Kyte & Doolittle (J. Mol. Biol.
157, 105-132 [1982] admit that "subjective adjustment" and "arbitrary
assignment" were used in producing their (now widely used) hydrophobicity scale.
There's no real reason why you can't use chain flexibility, alpha helix
propensity, percentage buried .... etc.  The amino acid substitution matrices
used in sequence alignment can be considered to integrate all these factors
without imposing the physical bias of a particular scala - just a more
evolutional bias.
Further, although quantification of the quality of sequence alignments is a
subject for great debate this should give some measure for the relatedness of
protein sequences.
	Criticism and discussion on the above greatly welcomed,
			Bernard


Bernard Murray, Ph.D.
bernard at elsie.nci.nih.gov (National Cancer Institute, NIH, Bethesda MD, USA)




More information about the Methods mailing list