On 19/10/2012 10:33, Dr. Josef Maier - IStLS wrote:
Hello Angus,
you could use preg with following pattern as written in regular
expressions:
([KRH][^DE][^DE][^DE])|([^DE][KRH][^DE][^DE])|([^DE][^DE][KRH][^DE])|([^DE][^DE][^DE][KRH])
Alternatively you could search with four different PROSITE-style
patterns using fuzzpro and combine the result tables:
[KRH]{DE}(3)
{DE}(1)[KRH]{DE}(2)
{DE}(2)[KRH]{DE}(1)
[KRH]{DE}(3)
You can also put the four patterns in a file and use the syntax -pattern
@patternfile
% cat patternfile
>first
[KRH]{DE}(3)
>second
{DE}(1)[KRH]{DE}(2)
>third
{DE}(2)[KRH]{DE}(1)
>fourth
[KRH]{DE}(3)
For searching with combinations of PROSITE patterns, amino acid
compositions and eventually AAINDEX profiles we had made a free web
application for the University of Oslo, the SAPA tool:
http://sapa-tool.uio.no/sapa/index.php
Interesting. I'll take a look.
E.g. searching a 4-letter subsequence with the PROSITE-style patterns
"[KRH].{DE}(4)", where the dot operator means logical AND, will produce
a list of all subsequences having the two patterns in that application.
Maybe the possibility to combine more than one PROSITE-style pattern
within a fuzzpro search with logical AND would be a useful extension for
fuzzpro improvement. Often more than one pattern is given for a domain
or functional site in the PROSITE pattern database. Of course preg will
do the job, however, the PROSITE patterns have to be rewritten as
regular expressions.
We also have a long standing offer to revive scrutineer, written by
Peter Sibbald at EMBL some years ago but it would need translation from
Pascal (not too hard to do). It loaded SwissProt into memory and had
interesting ways to search for motif patterns.
regards,
Peter Rice
EMBOSS Team
_______________________________________________
EMBOSS mailing list
EMBOSS@lists.open-bio.org
http://lists.open-bio.org/mailman/listinfo/emboss
