Fig. 1
From: A single-cell atlas of the murine limb skeleton integrating the developmental and adult stages
Flow diagram of computation methods. (1) Publicly available datasets were preprocessed using Cell Ranger and Scater. (2) Individual datasets were then clustered and annotated using Seurat v4. (3) We tested scVI, scANVI, scGen and Scanorama and (4) evaluated them with the scIB benchmarking package. scANVI was used to create the final Limb Skeletal Cell Atlas. (5) Cell-cell interactions for early limb bud signaling pathways were predicted using CellPhoneDB and (6) the growth plate was reconstructed with the use of Monocle3 pseudotemporal ordering. (7) Finally, an in silico knockout simulation for Sox9 was performed with Monocle3 and CellOracle and (8) evaluated with in vivo wild-type and knockout data integrated/projected with Harmony and scvi-tools. WT wild-type, KO knockout
