Disjoint PCA models for marker identification and classification of cancer types using gene expression data

Background The parallel monitoring of the expression profiles of thousands of genes seems particularly promising for a deeper understanding of cancer biology and to identify molecular signatures supporting the histological classification schemes of neoplastic specimens. However, molecular diagnostic based on microarray data presents major challenges due to the complex, multiclass nature and to the overwhelming number of variables characterizing gene expression databases of multiple tumor samples. Thus, the development of multiclass; classification schemes and of marker selection methods, that allow the simultaneous classification of multiple tumor types and the identification of those genes that are most likely to confer high classification accuracy, is of paramount importance.Methods. A computational procedure for marker identification and classification of multiclass gene expression data through the application of disjoint principal component models, based on the Soft independent Modeling of Class Analogy approach (SIMCA), is described. The identified features represent a rational and dimensionally reduced base for understanding the basic biology of diseases, defining targets of therapeutic intervention, and developing diagnostic tools for the identification and classification of multiple pathological states.Results. The method has been tested on 2 different microarray data sets obtained from various human tumor samples: i) acute leukemias, and ii) small round blue-cell tumors.Conclusions. The results demonstrate that the disjoint PCA modeling procedure allows the identification of specific phenotype markers and provides the assignment to multiple classes for previously unseen instances.