Variant filters using segregation information improve mapping of nectar-production genes in sunflower (Helianthus annuus L.).

Accurate variant calling is critical for identifying the genetic basis of complex traits, yet filters used in variant detection may inadvertently exclude valuable genetic information. In this study, we compare common sequencing depth filters, used to eliminate error-prone variants associated with repetitive regions and technical issues, with a biologically relevant filtering approach that targets expected Mendelian segregation. The resulting variant sets were evaluated in the context of nectar volume quantitative trait loci (QTL) mapping in sunflower (Helianthus annuus L.). Our previous research failed to detect an interval containing a strong candidate gene for nectar production (HaCWINV2). We removed hard filters and implemented a chi-square goodness-of-fit test to retain variants that segregate according to expected genetic ratios. We demonstrate that biologically relevant filtering retains more significant QTL and candidate genes, including HaCWINV2, while removing variants due to technical errors more effectively, and accounted for 48.55% of nectar production phenotypic variation. In finding nine putative homologs of Arabidopsis genes with nectary function within QTL regions, we demonstrate that this filtering strategy has a higher power of true variant detection in QTL mapping than the commonly used variant depth filtering strategy. Future research will adapt the technique to multiple population contexts, such as genomic selection.