Fig. 5

Consistency of multiomics clustering according to their donors. Bar plots (mean ± s.e.) of Adjusted Rand Index (ARI) values of multiomics clustering using different batch-effect correction algorithms in balanced and confounded scenarios. Three integrative tools were used, including SNF (a), intNMF (b), and iClusterBayes (c). Expression profiles from 36 samples from three donors derived from 12 batches in each omics type were randomly selected from the balanced and confounded datasets and further used to integrate cross-omics data. In order to eliminate selection bias, the random selection and cross-omics integration was conducted ten times. Mean value of the dataset without correction (raw) in each panel was plotted in dashed line. Performances between the raw group and BECA groups were compared using Student’s t-test. A group with the performance significantly higher than the raw group was marked with stars (*). Symbolic number coding of p-value was used as: *** (p ≤ 0.001), ** (0.001 < p ≤ 0.01), * (0.01 < p ≤ 0.05)