Abstract
BACKGROUND: Single-cell sequencing approaches allow gene expression to be measured at the single-cell level, providing opportunities and challenges to study the aetiology of complex diseases, including cancer. METHODS: Based on single-cell gene and lncRNA expression levels, we proposed a computational framework for cell type identification that fully considers cell dropout characteristics. First, we defined the dropout features of the cells and identified the dropout clusters. Second, we constructed a differential co-expression network and identified differential modules. Finally, we identified cell types based on the differential modules. RESULTS: The method was applied to single-cell melanoma data, and eight cell types were identified. Enrichment analysis of the candidate cell marker genes for the two key cell types showed that both key cell types were closely related to the physiological activities of the major histocompatibility complex (MHC); one key cell type was associated with mitosis-related activities, and the other with pathways related to ten diseases. CONCLUSIONS: Through identification and analysis of key melanoma-related cell types, we explored the molecular mechanism of melanoma, providing insight into melanoma research. Moreover, the candidate cell markers for the two key cell types are potential therapeutic targets for melanoma.