"Dr Engelbert Buxbaum" <engelbert_buxbaum at hotmail.com> wrote in message
news:d09481$hvj$01$1 at news.t-online.com...
>bioinformatically at gmail.com wrote:
>>> I'm not a bioinformatician but, I was wondering ... many sequence
>> alignments make comparisons at the amino-acid level because this
>> technique is more sensitive than using nucleotide bases. Right? What
>> are the problems, other than heuristic, concerned with increasing the
>> alphabet again wrt amino-acid 'environments' --- say, doublets or
>> triplets? Would the resulting information have improved utility in
>> structural or evolutionary analyses?
>> There was a paper on this in the J.Mol.Biol. in the early 90's (sorry,
> don't recall the exact reference). The idea was to put the frequency of
Paper by Marin van Heel
A new family of powerful multivariate statistical sequence analysis
J Mol Biol. 1991 Aug 20;220(4):877-87.
PMID: 1880802 [PubMed - indexed for MEDLINE]
> all aa triplets into a 3-d matrix, then do a matrix reduction and use
> the result to calculate the homology of proteins (avoiding the
> computationally expensive multi-sequence alignment).