Osteoblast differentiation is crucial for bone formation and maintaining skeletal integrity. Although it is now understood that this process exhibits significant heterogeneity across developmental stages and tissue microenvironments, the underlying mechanisms remain largely unexplored. In the present study, we introduceTrajAtlas, a comprehensive framework that addresses this gap in knowledge.TrajAtlascomprises four modules: a reference atlas (Differentiation Atlas), a differentiation model (Differentiation Model), a tool for differential pseudotime analysis (TrajDiff), and a method for pseudotemporal gene module detection (TRAVMap). By leveraging single-cell technologies,TrajAtlasoffers a systematic approach to exploring the multi-scale heterogeneity among cells, genes, and gene modules within population-level trajectories across diverse tissues and age groups. We systematically investigate the impact of age and injury on osteogenesis, providing new insights into osteoporosis and bone regeneration. In conclusion, our comprehensive framework offers novel insights into osteogenesis and provides a valuable resource for understanding the complexities of bone formation.Author SummaryOsteoblasts, the cells responsible for bone formation, can originate from various cellular sources. However, it’s unclear how different progenitor cells differentiate into osteoblasts, and how this process is influenced by factors such as age and tissue location. This knowledge gap stems from the lack of comprehensive databases and tools to decipher the differentiation process. In this study, we introduce TrajAtlas, a comprehensive framework designed to bridge this gap. To explore the cellular origins of osteoblasts, we constructed an atlas centered on osteogenesis. To answer how progenitor cells differentiate to osteoblasts, we developed a model that reveals the dynamic regulatory landscape during this process. To elucidate the influence of age and tissue location on differentiation, we built a tool for differential analysis. Furthermore, to identify conserved patterns of differentiation, we developed an approach to detect pseudotemporal gene modules. We validated the effectiveness of this framework by applying it to more datasets, unveiling novel cell states associated with injury. Notably, this framework focuses on dynamic processes, with the potential for broader applications in studying cell differentiation and complementing cell-centric analyses.