Page 190 - Plant Canada 2024 Proceeding

PLANT CANADA 2024

were characterized using nanopore reads, and panSV was developed. In addition, short-read DNA was
used for SNP analysis, and RNA sequence data was used for gene-annotation and expression analysis.

Results:
A pan-genome for 50 spring type canola parents was constructed using nanopore-based genome
assembly. This approach accurately assembled complex genomic regions and enhanced the quality of
the pan-genome. Structural variants were characterized using long-reads, and >230K SV were identified
across the 50 NAM parents, with almost 50% of them being present in close proximity to 5Kb gene
regions. Interestingly, 1% of the SV are expected to have evolved from transposable elements (especially
by active retro-transposons). This resource provides a valuable tool for understanding the genetic
architecture of a population and identifying genes and genetic variations associated with desirable traits.
This study reports the successful construction of a pan-genome for 50 spring NAM canola parents using
long-read based genome assembly and structural variant annotation. The pan-genome provides a
comprehensive representation of the genetic diversity present in the spring type canola population,
including rare and novel genomic variations not captured in traditional reference genomes. The resulting
pan genome resources will be a valuable resource for genetic studies and breeding efforts in canola,
ultimately leading to improved crop yield and quality.

*[O170] GENOME-WIDE ASSOCIATION AND GENOMIC SELECTION FOR OIL AND FATTY ACID
1
1
PROFILE IN RAPESEED (BRASSICA NAPUS L.). Jared Bento , Jia Sun , Sakaria Liban , Curt
1
McCartney , Harmeet Chawla , and Robert Duncan . Department of Plant Science, University of
1
1
1 1
Manitoba, 66 Dafoe Road, Winnipeg, MB, Canada, R3T2N2
Correspondence to: rob.duncan@umanitoba.ca

The overarching goals of rapeseed (Brassica napus L.) breeding efforts include the improvement of yield-
and seed-quality-related traits to meet the evolving demands of a growing population. Genome-wide
association studies (GWAS) and genomic selection (GS) are important biotechnological methods that
provide the potential to improve selection efficiency and shorten crop breeding cycles. These
biotechnologies facilitate more rapid responses to agronomic and quality challenges, as well as enhance
the sustainability of plant breeding programs by reducing the crop inputs, carbon emissions, and other
procedures associated with long-term programs.

This study has three main objectives: 1) GWAS to identify quantitative trait loci (QTL) for five seed quality
traits (overall oil content, erucic, oleic, linoleic, and linolenic acids), 2) evaluating GS accuracy in
predicting rapeseed hybrid fatty acid profile components, and 3) evaluating the "GS + de novo GWAS"
method proposed to improve GS prediction accuracy.
We analyzed 454 Brassica napus genotypes (92 parents, 362 hybrids) grown over 48 site-years. All
genotypes were genotyped via Brassica 60K Illumina SNP array.

Across 24 unique GWAS analyses, 161 QTL were identified, including 22 QTL for erucic acid. Several
QTL coincide with candidate genes identified in literature. Novel QTL have also been identified for all five
traits, warranting further candidate gene investigation.

Prediction accuracies for each seed quality trait were compared across 135 unique analyses, evaluating
responses to GS models (nine regression models), population (five model training/validation population
designs), and marker density (three marker sets containing low, intermediate, and high densities).
Prediction accuracies (represented by correlation between predicted and actual phenotypes) range from
0.023 (overall oil content) to 0.897% (linoleic acid content). Prediction accuracies exhibited negative
correlation to trait complexity, positive correlation to degree of training/validation population relatedness,
and no significant differences among marker densities or parametric models. Machine learning models
performed either equivalent or worse than common parametric models. Overall oil content, the most
complex trait analyzed, showed accuracy improvements as high as 0.745 when varying the
aforementioned factors.

By incorporating significant markers from GWAS, the accuracies of “GS + de novo GWAS” methods were
compared to conventional GS models. Prediction accuracy response appears trait-dependent: two traits

189