Defining early human developmental identity: A curated, cross-platform marker framework

Early human development involves dynamic cell-state transitions, complicating distinction between transient transcriptional changes and stable lineage commitment in human pluripotent stem cell models. To address this challenge, we establish a curated, cross-platform marker framework for pluripotency and early germ-layer identity. Starting from an unbiased RNA-seq discovery set, we validated candidates by qPCR and evaluated the panel across independent bulk RNA-seq and embryonic single-cell and spatial transcriptomic datasets. This yielded 67 validated markers, of which 66 were retained for downstream transcriptomic analyses. Our framework identified early developmental states across heterogeneous datasets and showed lineage-associated enrichment in human and macaque embryonic references despite substantial transcriptional variability. Protein-level concordance was observed for detectable targets in KOLF2.1J proteomic data. To enable broad application, we introduce DeepDiff, a web-based resource integrating the marker framework and enabling automated fate classification. Together, this work provides a transferable framework for assessing early developmental identity across experimental contexts.