Consensus Pituitary Atlas, a scalable resource for annotation, novel marker discovery and analyses in pituitary gland research

Previous single-cell profiling studies of the pituitary gland have yielded minimally reproducible insights largely due to their low statistical power and methodological inconsistencies. To address this problem, we generated a uniformly pre-processed Consensus Pituitary Atlas (CPA) using all existing mouse pituitary single-cell datasets (267 biological replicates, >1.1 million high-quality cells). The CPA revealed novel cell typing and lineage markers, including low-expression transcripts that previous analyses could not detect. The scale of the CPA enabled the development of machine learning models to automate and standardize cell type annotation and doublet identification for future studies. Leveraging the curated metadata, we identified sex-biased and age-dependent gene expression patterns at cell type resolution. To identify drivers of cell fates, first we determined consensus cell communication patterns. Secondly, we used RNA-sequencing and chromatin accessibility data to identify transcription factors associated with cell fates across modalities. The epitome platform acts as an interface with the CPA, allowing streamlined user-friendly analyses.