QTL mapping for heading date and yield-related traits in Korean japonica rice varieties and development of a candidate gene prioritization framework

Improving grain yield while maintaining regional adaptability in temperate japonica rice remains a key breeding challenge. Eight yield-related agronomic traits were quantitatively evaluated in the 185 F 8 recombinant inbred lines derived from crossing Korean cultivars Nampyeong and Samkwang under field and greenhouse conditions. A genetic linkage map harboring 610 markers (1,430.1 cM) was constructed using target capture sequencing and derived cleaved amplified polymorphic sequence (dCAPS) markers. QTL analysis identified 14 loci under field conditions and 10 under greenhouse conditions, with two QTLs detected under both conditions. For more effective prioritization of candidate genes within broad QTL intervals, we developed a computational framework integrating (1) variant impact, (2) positional information, (3) semantic similarity between QTL traits and gene annotations, and (4) semantic similarity between QTL traits and PubMed-derived information, with the two semantic similarity scores calculated using a sentence-transformer model. Key findings included the qHD7 locus, a major heading date QTL detected under both field and greenhouse conditions. The candidate gene for qHD7 was PRR37 , and the Samkwang-type PRR37 allele was associated with earlier heading and lower yield under field conditions, suggesting a possible trade-off between growth duration and productivity. In addition, OsUBR7 was prioritized as a plausible candidate for the major culm length QTL qCL6 , and PAY1 as a strong candidate for a yield-associated QTL on chromosome 8. These findings provide useful information for marker-assisted selection in temperate japonica rice and highlight the potential utility of the candidate gene prioritization framework as a complementary approach for identifying causal genes within QTL intervals.