For this data set we induced models for two different classification problems. With the first model we are trying to distinguish between diffuse and intestinal tumor gastric samples and normal gastric tissue.
Platform: Affymetrix GeneChip Human Full Length Array HuGeneFL
Diagnostic classes:
- normal gastric tissue (Normal): 8 examples (26.7%)
Number of genes: 4522 Number of samples: 30 Note: From the originally measured 7128 probe sets we removed genes that were not present (P) in at least one sample
Predictive accuracy with 10-fold cross validation (classifying using the best projection with eight attributes):
Following are the three best-ranked visualization with eight, six and four attributes in respect to the visualization score, that is, visualizations where examples from different diagnostic classes are best separated:
Score: 99.57% Genes: D26129_at: ribonuclease, RNase A family, 1 (pancreatic), RNASE1 U66052_at: Iduronate 2-sulfatase (Hunter syndrome), IDS X83416_s_at: prion protein (p27-30) (Creutzfeld-Jakob disease, Gerstmann-Strausler-Scheinker syndrome, fatal familial insomnia), PRNP U46006_s_at: cysteine and glycine-rich protein 2, CSRP2 U03105_at: proline-rich nuclear receptor coactivator 1, PNRC1 U50360_s_at: calcium/calmodulin-dependent protein kinase (CaM kinase) II gamma, CAMK2G X81817_at: B-cell receptor-associated protein 31, BCAP31 U07969_s_at: cadherin 17, LI cadherin (liver-intestine), CDH17
Score: 98.99% Genes: M62628_s_at: Hypothetical protein MGC27165, MGC27165 D87742_at: C219-reactive peptide /// AAAP6077 /// similar to C219-reactive peptide, KIAA0268 /// UNQ6077 /// LOC440751 U86755_s_at: a disintegrin and metalloproteinase domain 17 (tumor necrosis factor, alpha, converting enzyme), ADAM17 U03105_at: proline-rich nuclear receptor coactivator 1, PNRC1 U95040_at: tripartite motif-containing 28, TRIM28 D78132_s_at: Ras homolog enriched in brain, RHEB