Additional file 1 of geneBasis: an iterative approach for unsupervised selection of targeted gene panels from scRNA-seq

Additional file 1: Fig. S1. Schematic visualisation for cell neighbourhood preservation score workflow. Fig. S2. Systematic assessment of the ‘completeness’ of the generated gene panels. Fig. S3. SCMER introduces redundancy in the selections when selecting a small number of genes. Fig. S4. geneBasis is robust to initial selections and quickly finds missing sources of variation. Fig. S5. geneBasis accounts for batch effects even with highly unbalanced celltype composition. Fig. S6. Detailed analysis of the gene selection for mouse embryogenesis. Fig. S7. Detailed analysis of the selection for spleen and pancreas . Panels A-D correspond to spleen; panels E-G correspond to pancreas. Fig. S8. geneBasis selects genes that recover biological heterogeneity in seqFISH datasets. Fig. S9. Performing PCA for mapping manifolds with selections does not change cell neighborhood preservation score or gene prediction score. Fig. S10. Minkowski distance of order p=3 provides the resolution to select rare markers and genes expressed in a low fraction of cells.