Screening Discriminating SNPs for Chinese Indigenous Pig Breeds Identification Using a Random Forests Algorithm.

Area of Science:

• Animal Genetics
• Genomics
• Bioinformatics

Background:

• Chinese indigenous pig breeds possess significant genetic diversity but lack established identification methods.
• Effective breed identification is crucial for conservation and management of these unique genetic resources.

Purpose of the Study:

• To develop an efficient and accurate method for identifying Chinese indigenous pig breeds.
• To select a minimal set of highly discriminative single-nucleotide polymorphisms (SNPs) for breed identification.

Main Methods:

• Genotyping 1059 individuals from 18 Chinese indigenous and 5 cosmopolitan pig breeds using 62,822 SNPs.
• Applying linkage disequilibrium (LD) pruning and principal component analysis (PCA) for initial SNP filtering and population structure analysis.
• Utilizing the random forest (RF) algorithm with mean decreasing accuracy (MDA) to select an optimal panel of 1000 breed-discriminative SNPs.

Main Results:

• PCA revealed clear differentiation between Chinese indigenous and cosmopolitan breeds, with less variation among Yangtze River Delta indigenous breeds.
• The RF algorithm, optimized with 1000 trees, identified a panel of 1000 SNPs with high discriminative power.
• A machine learning classification model achieved over 99.3% accuracy in identifying breeds from a test set of 318 samples.

Conclusions:

• A low-density SNP panel combined with machine learning provides a highly accurate method for multi-breed identification of Chinese indigenous pigs.
• This approach facilitates efficient breed identification, supporting conservation efforts and genetic resource management.
• The developed method offers a practical tool for distinguishing diverse Chinese pig populations.