A simulation to analyze feature selection methods utilizing gene ontology for gene expression classification.

Gene expression profile classification is a pivotal research domain assisting in the transformation from traditional to personalized medicine. A major challenge associated with gene expression data classification is the small number of samples relative to the large number of genes. To address this problem, researchers have devised various feature selection algorithms to reduce the number of genes. Recent studies have been experimenting with the use of semantic similarity between genes in Gene Ontology (GO) as a method to improve feature selection. While there are few studies that discuss how to use GO for feature selection, there is no simulation study that addresses when to use GO-based feature selection. To investigate this, we developed a novel simulation, which generates binary class datasets, where the differentially expressed genes between two classes have some underlying relationship in GO. This allows us to investigate the effects of various factors such as the relative connectedness of the underlying genes in GO, the mean magnitude of separation between differentially expressed genes denoted by δ, and the number of training samples. Our simulation results suggest that the connectedness in GO of the differentially expressed genes for a biological condition is the primary factor for determining the efficacy of GO-based feature selection. In particular, as the connectedness of differentially expressed genes increases, the classification accuracy improvement increases. To quantify this notion of connectedness, we defined a measure called Biological Condition Annotation Level BCAL(G), where G is a graph of differentially expressed genes. Our main conclusions with respect to GO-based feature selection are the following: (1) it increases classification accuracy when BCAL(G) ≥ 0.696; (2) it decreases classification accuracy when BCAL(G) ≤ 0.389; (3) it provides marginal accuracy improvement when 0.389

Expression of the tumor suppressor Krüppel-like factor 4 as a prognostic predictor for colon cancer.

BACKGROUND: The zinc finger transcription factor Krüppel-like factor 4 (KLF4) regulates numerous physiologic processes, including proliferation, differentiation, and development. Studies also showed that KLF4 is involved in tumorigenesis and somatic cell reprogramming. Here, we aimed to assess whether KLF4 is a prognostic indicator for colon cancer. METHODS: Levels of KLF4 were measured by immunohistochemical analysis of a tissue microarray containing 367 independent colon cancer sections. Univariate data analysis was done in addition to construction of multivariate models with several clinicopathologic factors to evaluate KLF4 as an independent predictor of survival and cancer recurrence (disease-free survival). RESULTS: Colon cancer tissues had significantly overall lower KLF4 levels compared with noncancer tissues (P < 0.0001). Using logistic regression, a trend was noted for decreased odds of KLF4 expression in higher stages of tumors. In univariate and multivariate analyses, KLF4 was a significant predictor of survival and recurrence. CONCLUSIONS: KLF4 expression is significantly downregulated in colon cancer, and loss of KLF4 is an independent predictor of survival and recurrence. IMPACT: These findings suggest that KLF4 may serve as a prognostic biomarker for colon cancer.

Krüppel-like factor 5 is a crucial mediator of intestinal tumorigenesis in mice harboring combined ApcMin and KRASV12 mutations.

BACKGROUND: Both mutational inactivation of the adenomatous polyposis coli (APC) tumor suppressor gene and activation of the KRAS oncogene are implicated in the pathogenesis of colorectal cancer. Mice harboring a germline ApcMin mutation or intestine-specific expression of the KRASV12 gene have been developed. Both mouse strains develop spontaneous intestinal tumors, including adenoma and carcinoma, though at a different age. The zinc finger transcription factor Krüppel-like factor 5 (KLF5) has previously been shown to promote proliferation of intestinal epithelial cells and modulate intestinal tumorigenesis. Here we investigated the in vivo effect of Klf5 heterozygosity on the propensity of ApcMin/KRASV12 double transgenic mice to develop intestinal tumors. RESULTS: At 12 weeks of age, ApcMin/KRASV12 mice had three times as many intestinal tumors as ApcMin mice. This increase in tumor number was reduced by 92% in triple transgenic ApcMin/KRASV12/Klf5+/- mice. The reduction in tumor number in ApcMin/KRASV12/Klf5+/- mice was also statistically significant compared to ApcMin mice alone, with a 75% decrease. Compared with ApcMin/KRASV12, tumors from both ApcMin/KRASV12/Klf5+/- and ApcMin mice were smaller. In addition, tumors from ApcMin mice were more distally distributed in the intestine as contrasted by the more proximal distribution in ApcMin/KRASV12 and ApcMin/KRASV12/Klf5+/- mice. Klf5 levels in the normal-appearing intestinal mucosa were higher in both ApcMin and ApcMin/KRASV12 mice but were attenuated in ApcMin/KRASV12/Klf5+/- mice. The levels of beta-catenin, cyclin D1 and Ki-67 were also reduced in the normal-appearing intestinal mucosa of ApcMin/KRASV12/Klf5+/- mice when compared to ApcMin/KRASV12 mice. Levels of pMek and pErk1/2 were elevated in the normal-appearing mucosa of ApcMin/KRASV12 mice and modestly reduced in ApcMin/KRASV12/Klf5+/- mice. Tumor tissues displayed higher levels of both Klf5 and beta-catenin, irrespective of the mouse genotype from which tumors were derived. CONCLUSIONS: Results of the current study confirm the cumulative effect of Apc loss and oncogenic KRAS activation on intestinal tumorigenesis. The drastic reduction in tumor number and size due to Klf5 heterozygosity in ApcMin/KRASV12 mice indicate a critical function of KLF5 in modulating intestinal tumor initiation and progression.