Clustering of joint single-cell RNA-Seq (scRNA-Seq) data is often challenged by confounding factors, such as batch effects and biologically relevant variability. Existing batch effect removal methods typically require strong assumptions on the composition of cell populations being near identical across samples. University of Oxford researchers present CIDER, a meta-clustering workflow based on inter-group similarity measures. They demonstrate that CIDER outperforms other scRNA-Seq clustering methods and integration approaches in both simulated and real datasets. Moreover, the researchers show that CIDER can be used to assess the biological correctness of integration in real datasets, while it does not require the existence of prior cellular annotations.
CIDER can evaluate the biological accuracy of integration results
without reliance on ground truth
A The diagram shows two scenarios, one, where the cell types from two batches are correctly aligned, and one, where the two heterogeneous cell types (1 and 3) from two batches are falsely aligned, often because of overcorrection. B The schematic diagram elucidates the workflow for assessing the integrated outcome to identify falsely aligned clusters, which have lower inter-group similarity. CIDER can identify the second scenario without prior information of cell types