Machine learning-based classification of roses using 18 SNP markers for optimized genebank management
Laurine Patzer1, Marcus Linde1, Thomas Debener2
1Institute of Plant Genetics, Section Molecular Plant Breeding, Leibniz University Hannover, Herrenhäuser Straße 2, 30419, Hannover, Germany.
Plant Methods
|January 6, 2026
Summary
Machine learning with SNP markers accurately classifies rose cultivars, revealing genetic relationships and improving genebank management. This approach helps align traditional categories with genomic structure for better rose identification.
Area of Science:
- Botany
- Genetics
- Bioinformatics
Background:
- Rose cultivar classification is challenging due to complex breeding histories and genetic heterogeneity.
- Traditional horticultural categories often do not align with distinct genetic groups.
- Previous studies using molecular markers in roses are limited in scope.
Purpose of the Study:
- To evaluate the alignment of horticultural rose classifications with their genomic structure.
- To develop robust tools for managing large rose germplasm collections.
- To assess the effectiveness of machine learning in resolving rose genetic relationships.
Main Methods:
- Utilized 18 single nucleotide polymorphism (SNP) markers across 1,345 rose accessions.
- Applied multiple unsupervised and supervised machine learning algorithms for clustering and classification.
- Compared clustering results with traditional horticultural labels.
Main Results:
- Unsupervised clustering consistently identified genetically distinct groups, such as alba and damask roses.
- Supervised models achieved high classification accuracies (up to 100%) for complex hybrid groups like tea and bengal roses.
- Certain horticultural groups, including miniature and kordesii hybrids, clustered together despite differing labels, indicating shared genomic backgrounds.
Conclusions:
- Machine learning analysis of SNP data reliably resolves genetic relationships in rose cultivars.
- The findings provide insights into the concordance between horticultural classifications and genomic structure.
- Marker-based classification using machine learning is a valuable tool for genebank management, cultivar identification, and reassessing rose categories.
