Figure 1.

Data integrative analysis workflow. The schematic illustration depicts a typical data-driven analysis workflow and the data integrative approach for multi-study and multi-platform datasets developed here. NMI (Normalized Mutual Information) optimization is used to find an optimal trade-off between biological and technical differences for multi-study data. New data can be added to the same sample stratification based on remapping (applicable to t-SNE maps) or by matching clusters obtained in separate analysis using gene sets.