Locus minimization in breed prediction using artificial neural network approach

Area of Science:

• Animal genetics
• Bioinformatics
• Machine learning

Background:

• Molecular markers like microsatellites and SNPs are crucial for identifying animal breeds from small biological samples.
• Traditional methods face limitations including data unavailability, high costs, and complex analysis.

Purpose of the Study:

• To develop and demonstrate an artificial neural network (ANN) model for cost-effective and accurate breed identification.
• To create a webserver for accessible reference breed data, reducing the need for repeated genotyping.

Main Methods:

• Utilized microsatellite-based DNA fingerprinting data from 51,850 samples across 22 Indian goat breeds.
• Employed a multilayer perceptron model (a type of ANN) for locus minimization and breed identification.
• Developed a freely accessible webserver (http://nabg.iasri.res.in/bisgoat) for the research community.

Main Results:

• Achieved 96.63% training accuracy by minimizing the number of analyzed loci to nine.
• Demonstrated significant cost reduction through locus minimization.
• Established a web-based resource for reference breed data.

Conclusions:

• The ANN approach provides a cost-effective and accurate method for animal breed identification.
• The developed webserver serves as a valuable tool for identifying existing and new breeds, protecting intellectual property.
• This model offers a scalable solution for cost reduction in breed identification for various species, supporting conservation and improvement programs.