Skip to main content

Table 4 Performance comparison of com-abnormal gene selection in multi-view datasets

From: Robust hypergraph regularized non-negative matrix factorization for sample clustering and feature selection in multi-view gene expression data

Methods N Com-abnormal genes
PCA 25 KRT19, SPINK1, PRSS1, MUC6, VIM, HLA-A, SERPINA1, CTSB, KRT8, GNAS, ANXA2, HSPB1, HLA-C, KRT5, S100A6, PKM, HSP90AA1, ENO1, KRT17, MALAT1, COL1A1, ALDOA, LIPF, TMSB10, RPLP0
NMF 15 KRT19, SPINK1, PRSS1, HLA-A, SERPINA1, CTSB, KRT8, SPP1, GNAS, KRT5, S100A6, SERPINA3, COL1A1, TMSB10, RPLP0
GNMF 24 KRT19, PRSS1, MUC6, VIM, HLA-A, SERPINA1, CTSB, KRT8, GNAS, ANXA2, HSPB1, HLA-C, KRT5, S100A6, PKM, HSP90AA1, ENO1, KRT17, MALAT1, COL1A1, ALDOA, LIPF, TMSB10, RPLP0
NMFL2, 1 31 CEACAM5, KRT19, VIM, HLA-A, SERPINA1, CTSB, KRT8, CEACAM6, GNAS, ANXA2, HSPB1, HLA-C, KRT5, LAMC2, S100A6, ITGB1, PKM, HSP90AA1, ENO1, KRT17, MALAT1, MMP11, ITGB4, COL1A1, HSPG2, ALDOA, LDHA, LGALS3BP, S100A11, TMSB10, RPLP0
HNMF 32 CEACAM5, KRT19, VIM, HLA-A, SERPINA1, CTSB, KRT8, CEACAM6, GNAS, ANXA2, HSPB1, HLA-C, KRT5, S100A6, ITGB1, PKM, HSP90AA1, ENO1, S100A9, KRT17, LCN2, MALAT1, ITGB4, COL1A1, HSPG2, ALDOA, HSP90B1, LDHA, LGALS3BP, S100A11, TMSB10, RPLP0
SHNMF 31 CEACAM5, KRT19, VIM, HLA-A, SERPINA1, CTSB, KRT8, CEACAM6, GNAS, ANXA2, HSPB1, HLA-C, KRT5, S100A6, ITGB1, PKM, HSP90AA1, ENO1, S100A9, KRT17, LCN2, MALAT1, COL1A1, HSPG2, ALDOA, HSP90B1, LDHA, LGALS3BP, S100A11, TMSB10, RPLP0
RGNMF 33 EGFR, CCND1, KRT19, CD44, PRSS1, VIM, SLC2A1, CTSB, GNAS, ANXA2, HSPB1, HLA-C, KRT5, LAMC2, S100A6, ITGB1, PKM, HSP90AA1, ENO1, S100A9, H19, KRT17, ANXA1, MALAT1, ITGB4, COL1A1, ALDOA, HSPA1A, TNC, LDHA, LGALS3BP, S100A11, TMSB10
RHNMF 34 CEACAM5, KRT19, VIM, HLA-A, SERPINA1, CTSB, KRT8, CEACAM6, SPP1, GNAS, ANXA2, HSPB1, HLA-C, KRT5, KRT18, S100A6, ITGB1, PKM, HSP90AA1, ENO1, KRT17, HSPA5, LCN2, MALAT1, ITGB4, COL1A1, HSPG2, ALDOA, HSP90B1, LDHA, LGALS3BP, S100A11, TMSB10, RPLP0
  1. Note: Bold genes denote that they are selected simultaneously by these eight methods. Underlined genes denote that they can be selected by RHNMF. N represents the number of com-abnormal genes selected for every method