MPsrch
------
(C) Biocomputing Research Unit, University of Edinburgh, 1992
Smith and Waterman Protein Sequence Database searches on a MasPar
MPsrch (by Shane Sturrock and John Collins), the fastest implementation
of the exhaustive Best Local Similarity algorithm for Sequence Database
searching, is now available through BLITZ, a new e-mail service from the
EMBL. MPsrch runs on the 4096-processor MasPar MP-1 machine at the EMBL,
using the Smith & Waterman algorithm, performing up to 140,000,000 cell
updates per second.
Typical search times on the MasPar against Swiss-Prot 23 are:
5037 residues RYNR_RABIT 340 seconds (134,000,000 cell updates/second)
377 " ACTS_HUMAN 40 seconds ( 85,000,000 cell updates/second)
To get current help information, e-mail the word HELP in a mail message to
the Internet address:
Blitz at EMBL-Heidelberg.DE
To run a search, using default weight matrix and INDEL cost settings, try
sending an e-mail message with the word SEQ on the first line and the
sequence in free format on succeeding lines. The end of the sequence is
determined either by the end of your mail message, if you do not use an
automatic mail signature, or by inserting a new line beginning with the
word END.
For example:
SEQ
STKKKPLTQE QLEDARRLKA IYEKKKNELG LSQESVADKM GMGQSGVGAL
FNGINALNAY NAALLAKILK VSVEEFSPSI AREIYEMYEA VSMQPSLRSE
YEYPVFSHVQ AGMFSPELRT FTKGDAERWV STTKKASDSA FWLEVEGNSM
TAPTGSKPSF PDGMLILVDP EQAVEPGDFC IARLGGDEFT FKKLIRDSGQ
VFLQPLNPQY PMIPCNESCS VVGKVIASQW PEETFG
END
The Help information explains how you can specify a PAM matrix between 1
and 500 PAM's, an INDEL cost and the number of alignments to be returned
in the output.
The sensitivity of MPsrch depends on the choice of parameters, and
searches with default parameter settings alone may not always find the
most interesting results. Even with the same query sequence, one set of
parameters may find strong alignments best, while a different setting may
show up weaker relationships better. Because the search is so fast, we would
encourage you to experiment with your own choice of parameter settings.
In future developments, we hope to provide searches of other databases,
including a nucleotide search facility.
John Collins & Shane Sturrock, Des Higgins (Data Library)
Biocomputing Research Unit, & Roy Omond (Computer Group)
University of Edinburgh, EMBL,
Scotland, Heidelberg,
U.K. Germany.
