Prediction of the number of activated genes in multiple independent Cd(+2)- and As(+3)-induced malignant transformations of human urothelial cells (UROtsa)

• Published in: PloS one Vol. 9; no. 1; p. e85614
• Main Authors: Garrett, Scott H, Somji, Seema, Sens, Donald A, Zhang, Ke K
• Format: Journal Article
• Language: English
• Published: United States Public Library of Science (PLoS) 2014
• Subjects: Algorithms; Arsenic - pharmacology; Cadmium - pharmacology; Cell Line; Cell Transformation, Neoplastic - genetics; Cluster Analysis; Computational Biology - methods; Computer Simulation; Epithelial Cells - drug effects; Epithelial Cells - metabolism; Humans; Models, Genetic; Oligonucleotide Array Sequence Analysis; Reverse Transcriptase Polymerase Chain Reaction; Transcriptome - drug effects; Transcriptome - genetics; Urothelium - cytology; Urothelium - drug effects; Urothelium - metabolism
• ISSN: 1932-6203; 1932-6203
• DOI: 10.1371/journal.pone.0085614

Many toxic environmental agents such as cadmium and arsenic are found to be epidemiologically linked to increasing rates of cancers. In vitro studies show that these toxic agents induced malignant transformation in human cells. It is not clear whether such malignant transformation induced by a single toxic agent is driven by a common set of genes. Although the advancement of high-throughput technology has facilitated the profiling of global gene expression, it remains a question whether the sample size is sufficient to identify this common driver gene set. We have developed a statistical method, SOFLR, to predict the number of common activated genes using a limited number of microarray samples. We conducted two case studies, cadmium and arsenic transformed human urothelial cells. Our method is able to precisely predict the number of stably induced and repressed genes and the number of samples to identify the common activated genes. The number of independent transformed isolates required for identifying the common activated genes is also estimated. The simulation study further validated our method and identified the important parameters in this analysis. Our method predicts the degree of similarity and diversity during cell malignant transformation by a single toxic agent. The results of our case studies imply the existence of common driver and passenger gene sets in toxin-induced transformation. Using a pilot study with small sample size, this method also helps microarray experimental design by determining the number of samples required to identify the common activated gene set.