PRPS-ST: A protocol-agnostic self-training method for gene expression-based classification of blood cancers.

Gene expression classifiers are gaining increasing popularity for stratifying tumors into subgroups with distinct biological features. A fundamental limitation shared by current classifiers is the requirement for comparable training and testing data sets. Here, we describe a self-training implementation of our p robability r atio-based classification p rediction s core method (PRPS-ST), which facilitates the porting of existing classification models to other gene expression data sets. In comparison to gold standards, we demonstrate favorable performance of PRPS-ST in gene expression-based classification of DLBCL and B-ALL using a diverse variety of gene expression data types and pre-processing methods, including in classifications with a high degree of class imbalance. Tumors classified by our method were significantly enriched for prototypical genetic features of their respective subgroups. Interestingly, this included cases that were unclassifiable by established methods, implying the potential enhanced sensitivity of PRPS-ST.
Competing Interests: Disclosures: DWS and RDM are named inventors on patents for methods of classification of DLBCL. One of the patents that names DWS as an inventor is licensed to NanoString. One of the patents that names RDM as an inventor is licensed to Epizyme. Outside the submitted work, DWS also received personal consulting fees from Abbvie, AstraZeneca, Celgene, and Janssen, as well as research funding from NanoString and Roche. RDM reports personal fees from Celgene outside the submitted work. All other authors declare no competing interests.