Fig. 9

Performance evaluation results of scGAL on simulated datasets. The baseline model only uses the single-cell copy number data to cluster cells. For each configuration of the parameters used for simulating scRNA-seq data, the evaluation is conducted three times on SA501 dataset by using different seeds. A Evaluation results on datasets with different numbers of scRNA-seq cells. The results show scGAL performs robustly on different-sized datasets and surpasses the baseline model by a large margin. B Evaluation results on datasets generated with different values of the de.prob parameter of Splatter tool. The results showcase scGAL’s high robustness against the distribution changes of scRNA-seq data, and also performs much better than the baseline model