Profile Hidden Markov Models
Profile
hidden Markov models (HMMs) are statistical descriptions of sequence
conservation from multiple sequence alignments, and have been shown
to outperform standard pairwise sequence comparison methods, both
in terms of sensitivity and specificity. HMMs form the basis of
protein family and domain description libraries such as
SUPERFAMILY
and
Pfam. Kinomer v.1.0 is a multi-level library of HMMs where each
protein kinase group is represented by a number of HMMs, thereby
achieving great sensitivity and specificity for searching and classifying
protein kinases automatically into groups.
Running Kinomer locally
Sequences can be searched against the Kinomer library using the
hmmpfam program from the
HMM software suite HMMer. This is freely available for download from
http://hmmer.janelia.org. HMMer can be readily installed on a variety of Linux and UNIX platforms (including
Mac OS/X). The user manual , which is
also available from
the HMMer web site, will guide you through
the process of running your HMM searches locally.
The Kinomer v.1.0
HMM library is available for
download from this site.