An integrated reference atlas of human skeletal muscle.

BACKGROUND: Single-cell and single-nucleus RNA sequencing have transformed our understanding of human skeletal muscle biology, yet reproducibility and cross-study comparison remain limited by the lack of a unified reference framework and consistent cell-type annotation. METHODS: We systematically searched for scRNA-seq and snRNA-seq datasets from adult human skeletal muscle. Seven eligible studies were retrieved and harmonised. We benchmarked multiple integration strategies to construct a joint reference atlas and derived modality-aware marker panels. Selected findings were validated by immunofluorescence in muscle biopsies. FINDINGS: We generated a harmonised atlas comprising 122,000 cells and 630,000 nuclei from 88 healthy individuals and resolved 17 major skeletal muscle cell populations, spanning mononuclear compartments and multinucleated myofibers. Cross-modality analysis identified tissue- and modality-aware marker panels and nominated both established and previously unrecognised markers. NOVA1 emerged as a selective marker of fibro-adipogenic progenitors and was validated at the transcript and protein levels. Focusing on myonuclei, pseudotime modelling reconstructed differentiation trajectories from quiescent muscle stem cells to mature type I and type II myofibers and revealed lineage-specific programs, including transient activation of protocadherin-γ genes during type I myofiber differentiation. We further provide an interactive web application for marker-based cell-type prediction using the reference atlas. INTERPRETATION: This integrated reference atlas and accompanying annotation tool establish a standardised framework for human muscle transcriptomics, promoting consistent cell-type assignment and providing a baseline for future studies of muscle development, ageing, and disease. FUNDING: Else Kröner-Fresenius-Stiftung and the German Research Foundation.