Analysis of alcoholism data using support vector machines.

Area of Science:

• Genetics
• Bioinformatics
• Machine Learning

Background:

• The Collaborative Study on the Genetics of Alcoholism (COGA) dataset provides valuable genetic information.
• Understanding genetic contributions to alcoholism is crucial for developing targeted interventions.

Purpose of the Study:

• To apply supervised learning, specifically support vector machines (SVM), for analyzing genetic markers associated with alcoholism phenotypes.
• To identify significant marker sets and reduce dataset size using SVM with polynomial kernels.

Main Methods:

• Utilized support vector machine (SVM) with various polynomial kernel functions.
• Analyzed microsatellite marker data from autosomal chromosomes against twelve binary alcoholism phenotypes.
• Employed random genome region division and fourfold cross-validation for model assessment.

Main Results:

• Successfully identified associations between specific marker sets and selected phenotypes.
• Demonstrated high classification correctness, with predictions reaching 96% accuracy in cross-validation.
• Achieved dataset size reduction through feature selection guided by SVM.

Conclusions:

• Support vector machines are effective tools for identifying genetic associations in complex traits like alcoholism.
• The method shows promise for reducing dimensionality in large genetic datasets.
• Further validation with larger sample sizes is necessary to confirm predictive capabilities on new data.