Fig. 2: Clustering Accuracy on Simulated Multi-Omics Datasets.
From: GAUDI: interpretable multi-omics integration with UMAP embeddings and density-based clustering
This figure displays boxplots illustrating the Jaccard Index (JI) scores for clusters identified by various multi-omics integration methods compared to the ground-truth clusters in the simulated data (in these boxplots, the center line represents the median, box boundaries show the interquartile range (25th–75th percentiles), whiskers extend to points within 1.5 × IQR, and points beyond the whiskers indicate outliers). The analysis encompasses scenarios with 5, 10, and 15 pre-defined clusters. For each method, we present results for both heterogeneous (HET) and homogeneous (EQ) cluster distributions. The analysis is based on datasets comprising 500 samples, and the depicted results are aggregated over 1000 independent iterations of k-means clustering, ensuring robust and reliable performance evaluation. Source data are provided as a Source Data file.
