Coralysis enables sensitive identification of imbalanced cell types and states in single-cell data via multi-level integration.

Complex single-cell analyses now routinely integrate multiple datasets, followed by cell-type annotation and differential expression analysis. Current state-of-the-art integration methods often struggle with imbalanced cell types across datasets particularly when highly similar but distinct cell types are not present in all datasets. Inaccurate integration leads to incorrect annotations, affecting downstream analyses such as differential expression. To streamline single-cell data analysis, we introduce Coralysis, an all-in-one package featuring a sensitive integration algorithm, reference-mapping for accurate automatic annotation, and fine-grained cell-state identification. We demonstrate that Coralysis shows consistently high performance across diverse integration tasks, outperforming state-of-the-art methods particularly in challenging settings when similar cell types are imbalanced or missing. It accurately predicts cell-type identities across various annotation scenarios. A key strength of Coralysis is its ability to provide cell-specific probability scores, enabling the identification of transient and stable cell-states, along with their differential expression patterns. Importantly, Coralysis performs robustly on different types of single-cell data from transcriptomics to proteomics. Overall, Coralysis includes all the main steps of single-cell data analysis; it preserves subtle biological variation by improving the integration and annotation of imbalanced cell types, and identifies fine-grained cell-states-enabling a faithful analysis of the cellular landscape in complex single-cell experiments.