Bin Li and Warren J. Gallin Department of Biological Sciences University of Alberta

3
Dec 7, 2003 Poster Prediction of Half Activation Voltages of Voltage-gated Potassium Channels Based on Amino Acid Sequences Using Machine Learning Bin Li and Warren J. Gallin Department of Biological Sciences University of Alberta Edmonton, Canada

description

Prediction of Half Activation Voltages of Voltage-gated Potassium Channels Based on Amino Acid Sequences Using Machine Learning. Bin Li and Warren J. Gallin Department of Biological Sciences University of Alberta Edmonton, Canada. VKC. Data Processing. Basic Learning. Training data: - PowerPoint PPT Presentation

Transcript of Bin Li and Warren J. Gallin Department of Biological Sciences University of Alberta

Page 1: Bin Li and Warren J. Gallin Department of Biological Sciences University of Alberta

Dec 7, 2003

Poster

Prediction of Half Activation Voltages of Voltage-gated Potassium Channels Based on Amino Acid

Sequences Using Machine Learning

Bin Li and Warren J. GallinDepartment of Biological Sciences

University of Alberta Edmonton, Canada

Page 2: Bin Li and Warren J. Gallin Department of Biological Sciences University of Alberta

Dec 7, 2003

Poster

Basic Learning

Wrapper

Filter

Training data:58 VKC sequences -296 residues (features) eachClass: published Va values

Outlier Selection

ComparisonMatrix

KNN classifier

http://vkcdb.biology.ualberta.ca

Data Processing

Training set Construction of classifier

VKC

Page 3: Bin Li and Warren J. Gallin Department of Biological Sciences University of Alberta

Dec 7, 2003

Poster

Mathematics Biology?

KNN classifier

Feature selection

wrapper

BLOSUM62

Outlier selection

**

*

* **

*

Q/E

R/K

A/S/V

Machine learning may not provide definitive answers to biological problems,but it help propose newhypotheses for experimentaltests.