Utilization of genetic data can improve the prediction of type 2 diabetes incidence in a Swedish cohort.

The aim of this study was to measure the impact of genetic data in improving the prediction of type 2 diabetes (T2D) in the Malmö Diet and Cancer Study cohort. The current study was performed in 3,426 Swedish individuals and utilizes of a set of genetic and environmental risk data. We first validated our environmental risk model by comparing it to both the Finnish Diabetes Risk Score and the T2D risk model derived from the Framingham Offspring Study. The area under the curve (AUC) for our environmental model was 0.72 [95% CI, 0.69-0.74], which was significantly better than both the Finnish (0.64 [95% CI, 0.61-0.66], p-value < 1 x 10-4) and Framingham (0.69 [95% CI, 0.66-0.71], p-value = 0.0017) risk scores. We then verified that the genetic data has a statistically significant positive correlation with incidence of T2D in the studied population. We also verified that adding genetic data slightly but statistically increased the AUC of a model based only on environmental risk factors (RFs, AUC shift +1.0% from 0.72 to 0.73, p-value = 0.042). To study the dependence of the results on the environmental RFs, we divided the population into two equally sized risk groups based only on their environmental risk and repeated the same analysis within each subpopulation. While there is a statistically significant positive correlation between the genetic data and incidence of T2D in both environmental risk categories, the positive shift in the AUC remains statistically significant only in the category with the lower environmental risk. These results demonstrate that genetic data can be used to increase the accuracy of T2D prediction. Also, the data suggests that genetic data is more valuable in improving T2D prediction in populations with lower environmental risk. This suggests that the impact of genetic data depends on the environmental risk of the studied population and thus genetic association studies should be performed in light of the underlying environmental risk of the population.

Fast and Scalable Feature Selection for Gene Expression Data Using Hilbert-Schmidt Independence Criterion.

GOAL: In computational biology, selecting a small subset of informative genes from microarray data continues to be a challenge due to the presence of thousands of genes. This paper aims at quantifying the dependence between gene expression data and the response variables and to identifying a subset of the most informative genes using a fast and scalable multivariate algorithm. METHODS: A novel algorithm for feature selection from gene expression data was developed. The algorithm was based on the Hilbert-Schmidt independence criterion (HSIC), and was partly motivated by singular value decomposition (SVD). RESULTS: The algorithm is computationally fast and scalable to large datasets. Moreover, it can be applied to problems with any type of response variables including, biclass, multiclass, and continuous response variables. The performance of the proposed algorithm in terms of accuracy, stability of the selected genes, speed, and scalability was evaluated using both synthetic and real-world datasets. The simulation results demonstrated that the proposed algorithm effectively and efficiently extracted stable genes with high predictive capability, in particular for datasets with multiclass response variables. CONCLUSION/SIGNIFICANCE: The proposed method does not require the whole microarray dataset to be stored in memory, and thus can easily be scaled to large datasets. This capability is an important attribute in big data analytics, where data can be large and massively distributed.

Investigating the link between molecular subtypes of glioblastoma, epithelial-mesenchymal transition, and CD133 cell surface protein.

In this manuscript, we use genetic data to provide a three-faceted analysis on the links between molecular subclasses of glioblastoma, epithelial-to-mesenchymal transition (EMT) and CD133 cell surface protein. The contribution of this paper is three-fold: First, we use a newly identified signature for epithelial-to-mesenchymal transition in human mammary epithelial cells, and demonstrate that genes in this signature have significant overlap with genes differentially expressed in all known GBM subtypes. However, the overlap between genes up regulated in the mesenchymal subtype of GBM and in the EMT signature was more significant than other GBM subtypes. Second, we provide evidence that there is a negative correlation between the genetic signature of EMT and that of CD133 cell surface protein, a putative marker for neural stem cells. Third, we study the correlation between GBM molecular subtypes and the genetic signature of CD133 cell surface protein. We demonstrate that the mesenchymal and neural subtypes of GBM have the strongest correlations with the CD133 genetic signature. While the mesenchymal subtype of GBM displays similarity with the signatures of both EMT and CD133, it also exhibits some differences with each of these signatures that are partly due to the fact that the signatures of EMT and CD133 are inversely related to each other. Taken together these data shed light on the role of the mesenchymal transition and neural stem cells, and their mutual interaction, in molecular subtypes of glioblastoma multiforme.