Machine learning-based gait classification and genome-wide association identify a QTL for gait type in Colombian paso...

Area of Science:

• Genomics and Animal Genetics
• Locomotion and Biomechanics
• Computational Biology and Machine Learning

Background:

• Coordinated mammalian locomotion involves spinal circuits, with DMRT3 gene influencing strides and alternative gaits.
• The genetic basis for specialized gaits, such as the Colombian trocha and Colombian trot in paso horses, remains unexplained by DMRT3.
• Understanding the genetic architecture of complex locomotor traits is crucial for breed characterization and conservation.

Purpose of the Study:

• To identify genetic factors underlying the distinct gaits of the Colombian paso horse breed.
• To investigate the genetic differentiation responsible for the Colombian trocha and Colombian trot.
• To leverage artificial intelligence (AI) and machine learning for precise phenotyping in equine genomics.

Main Methods:

• Utilized inertial sensors and machine learning algorithms for accurate phenotyping of equine gaits (n=225 horses).
• Performed genome-wide association analysis (GWAS) on 85 horses using a 670K array.
• Identified and characterized quantitative trait loci (QTLs) and specific haplotypes associated with gait variations.

Main Results:

• A significant 2.43 Mb quantitative trait locus (QTL) was identified on chromosome ECA16.
• Lead variants within the QTL, rs1147402472 and rs1136628503, explained 48.6% of gait variance.
• A unique Colombian trot-specific haplotype, absent in other breeds, was discovered, indicating fine-scale genetic differentiation.

Conclusions:

• Fine-scale genetic differentiation at ECA16 underlies neural adaptations distinguishing complex equine locomotor traits.
• The identified QTL and associated genes (LHFPL4, SRGAP3, ATP2B2) provide insights into neurodevelopment and muscle regulation impacting gait.
• AI-assisted phenotyping is a powerful tool for genomic studies of complex traits in animal populations.