文档介绍：Hidden Mark o v Mo dels putational Biology
Applications to Protein Mo deling
UCSC
CRL
y y x
Anders Krogh
Mic hael Bro wn
I
Saira Mian
y y
Kimmen Sj
olander
Da vid Haussler
puter and Information Sciences
x Sinsheimer Lab oratories
Univ ersit y of California
San ta Cruz
CA
USA
email
krogh
nordig
ei
dth
dk
haussler
cse
ucsc
edu
August
Keyw ords
Hidden Mark o vModels
Multiple Sequence Alignmen ts
Database Searc hing
Globin
Kinase
EF
hand
EM algorithm
Abstract
Hidden Mark o v Mo dels
HMMs
are applied to the problems of statistical mo del
ing
database searc hing and m ultiple sequence alignmen t of protein families and protein
domains
These metho ds are demonstrated on the globin family
the protein kinase
catalytic domain
and the EF
hand calcium binding motif
In eac h case the parame
ters of an HMM are estimated from a training set of unaligned sequences
After the
HMM is built
it is used to obtain a m ultiple alignmen t of all the training sequences
It is also used to searc h the SWISS
PR OT
database for other sequences that are
mem b ers of the giv en protein family
orcon tain the giv en domain
The HMM pro duces
m ultiple alignmen ts of go o d qualit y that agree closely with the alignmen ts pro duced b y
programs that incorp orate three
dimensional structural information
When emplo y ed
in discrimination tests
b y examining ho w closely the sequences in a database
t the
globin
kinase and EF
hand HMMs
the HMM is able to distinguish mem b ers of these
families from non
mem b ers with a high degree of accuracy
Both the HMM and PR O
FILESEAR CH
a tec hnique used to searc h for relationships b et w een a protein sequence
and m ultiply aligned sequences
p erform b etter in these tests than PR OSITE
a dictio
nary of sites and patterns in proteins
The HMM app ears to ha v e a sligh t adv an tage
Presen t address
Electronics Institu