Endocrine Abstracts

Clinical exome and genome sequencing have substantially advanced the diagnosis of differences of sex development (DSD). However, many cases remain unsolved, in part due to the limited functional annotation of genes and–critically–non-coding regulatory elements that control sex-specific developmental programs. Current gene prioritization approaches do not adequately capture the cellular and temporal complexity of gonadal development, hindering the interpretation of both coding and non-coding variants. Here, we present single-cell tissue-specific gene prioritization using machine learning (STIGMA), a framework that leverages single-cell RNA sequencing (scRNA-seq) data to prioritize candidate genes and regulatory elements in a developmental context. STIGMA models the spatiotemporal dynamics of gene expression across cell types during gonadal differentiation, enabling the identification of genes and regulatory regions that are active in sex-specific lineages. Applying STIGMA to single-cell datasets of developing gonadal tissues, we demonstrate its ability to resolve cell-type-specific expression programs underlying testis and ovary development. Integrating these profiles with genomic data from individuals with DSD allows prioritization of both coding variants and non-coding variants affecting key regulatory networks. This approach highlights candidate genes and regulatory elements acting in supporting cells, germ cells, and steroidogenic lineages, thereby refining variant interpretation beyond conventional gene-centric methods. Overall, STIGMA provides a framework to link genetic variation to disrupted developmental trajectories in DSD. By capturing the cellular and temporal specificity of gene regulation, this approach improves the identification of causal genes and non-coding regulatory variants, offering new insights into the molecular basis of human sex development and its disorders.