HingeSeek uses Kernel Logistic Regression (KLR) to identify residues in a protein sequence that may be part of a hinge-bending region. A sequence can be either entered manually or uploaded in plain text or FASTA format. The full method will be described in a forthcoming paper.

HingeSeek is trained using data from the DynDom database.

Text input:

Or, upload sequence:

Hinge assignment threshold:
Residues with a predicted outcome at or below
this value will be classified as a hinge.

Sequence identity threshold:
Models trained on sequences above this sequence
identity will not be used to perform regression. The
default value 100 will use all training data.
This option increases the runtime to
around 2-3 minutes.