[BiO BB] How many peptide residues can form a meaningful/functional protein sequence?
idoerg at burnham.org
Tue Feb 11 21:49:58 EST 2003
It is actually almost meaningless to infer homology by sequence
identity. You may quite safely infer homology for proteins longer than
80 aa with a >30% pairwise identity, but there are quite a few homologs
with a seqid below that threshold, and for proteins shorter than 80, the
%ID is a bad estimator to homology, even as a rough yardstick. More
sophisticated statistics should be used, like Monte-Carlo simulations of
your alignment procedure. I.e., how good does your alignment score when
compared to the mean of a collection of random alignnment scores for
proteins of the same length and composition?
My favorite is to use blast2 (on NCBI's site) for two peptides. They
work out the statistics quite well.
For further reading look to:
Twilight zone of protein sequence alignments.
Protein Eng. 1999 Feb;12(2):85-94.
and references therein.
Iddo Friedberg, Ph.D.
The Burnham Institute
10901 N. Torrey Pines Rd.
La Jolla, CA 92037
Tel: +1 (858) 646 3100 x3516
Fax: +1 (858) 646 3171
#JIA YIYU# wrote:
> Hi all,
> Thanks James for his helps. Now I know at least one decapeptide sequence can affect the function of one particular protein.
> My motivation to ask my last question is about protein sequence homologous alignment because I doubt
> whether it is meanful action to align two protein sequence under a very low identity threshhold value for investigating the
> homologue between them. Clearly, if we have two protein sequences, both of which contain ten amino acid residues.
> But if the shortest length of one functionable peptide sequence was 4, then it is meanless
> to investigate the 30% homologue alignment between this two ten length peptide sequences.
> Maybe my question is naive. But further teaching is appreciated very much!
> Jia Yiyu
> BiO_Bulletin_Board maillist - BiO_Bulletin_Board at bioinformatics.org
More information about the BBB