Supplementary material

Recent advances in linking the human genome to traits have revealed that genetic variants affect nearly all human traits. Accurate phenotype measurement is crucial for successful genotype-phenotype analysis, but extracting complex morphological traits, such as the human face, remains challenging due to their multivariate and multipartite nature. In this study, we introduce a novel, data-driven, optimization-based phenotyping approach for genetic research. Unlike traditional methods that rely on selection, our approach emphasizes optimization. This method first maps out a detailed range of shape variations and then optimizes for traits with strong genetic relevance. Our findings demonstrate that this optimized method identifies genetic variants associated to facial shape more effectively than conventional techniques. Overall, this approach provides a promising alternative to non-optimized phenotyping methods and shows potential for broader application in genotype-phenotype analysis involving complex traits.