Popular Vote (popV, https://github.com/YosefLab/popV) is an ensemble of prediction models with an ontology-based voting scheme. PopV takes a consensus of experts approach to the task of automated cell type annotation. The input is an unannotated query data set together with an annotated reference data set. Both data sets are expected to contain raw count data and demonstrate that popV can be applied to UMI, as well as non-UMI-based technologies. PopV then runs eight different annotation methods: random forest (RF), support vector machine (SVM), scANVI, OnClass, Celltypist and k nearest neighbors (kNN) after batch correction with three single cell harmonization methods: scVI, BBKNN, and Scanorama. The eight prediction algorithms were chosen because they were shown to have good prediction accuracy and/or good harmonization performances. The notebook linked bellow provides a step by step tutorial on how to annotate new data using PopV, and the underlying variety of annotation methods, using the Tabula Sapiens dataset as the reference. For details on the methods implemented, and for further explanation about PopV, please refer to the publication.