Fig. 4
From: Contrastive self-supervised clustering of scRNA-seq data
Dataset-level analysis of real scRNA-seq data on ARI (Adjusted Rand Index) scores. The results aggregate 3 consecutive runs of all 13 explored methods over the 15 biological datasets. The dataset annotations (e.g. #1, #6) indicate the ranking of contrative-sc with KMeans and Leiden on each analyzed dataset: the best performing method 6 datasets and the second best on another 6