Identifying the biologically relevant gene categories based on gene expression and biological data: an example on prostate cancer

MOTIVATION

Most gene-expression based studies aim to identify genes with the capability of distinguishing different phenotypes. Although analysis at the genomic level is important, results of the molecular/cellular level are essential for understanding biological mechanisms. To deliver molecular/cellular-level results, a two-stage scheme is widely employed. This scheme just evaluates biological processes/molecular activities individually, totally overlooking the relationship between processes/activities. This treatment conflicts with the fact that most biological processes/molecular activities do not work alone. In order to deliver improved results, this shortcoming should be addressed.

RESULTS

We design a selection model from a novel perspective to directly detect important gene functional categories (each category represents a cellular process or a molecular activity). More importantly, the correlations between gene categories are considered. Contributed by this capability, the proposed method shows its advantages over others.

AVAILABILITY

the source code in Matlab is accessible via http://www.ee.cityu.edu.hk/~twschow/category_selection/category_selection.htm

Association of variants in theIL12Bgene with leprosy and tuberculosis

There is a great range in outcomes after mycobacterial infections, and this is probably due to individual variation in immune responses. One of the key cytokine regulators of the immune response is interleukin (IL-) 12. The IL12B gene encodes the p40 chain of both IL-12 and IL-23 and it has two major variant sites at which different alleles are associated with increased levels of gene expression and with susceptibility to a range of immune-related diseases. We hypothesized that IL12B variants associated with increased expression would be as associated with susceptibility to persistent mycobacterial infection. We tested this hypothesis by genotyping Indian subjects, having either leprosy or tuberculosis (TB), as well as ethnically matched controls. Subjects with leprosy were less likely to have the 3'UTR genotype associated with lower IL12B expression (P= 0.001). Subjects with TB were not only more likely to have the high-expressing IL12B promoter genotype (P= 0.01) but also more likely to have this in the same haplotype with the high expressing 3'UTR allele (P= 0.0009). These results suggest these infectious diseases may be improved by modulating IL-l2p40 production.

Feature extraction using recursive cluster-based linear discriminant with application to face recognition

A novel recursive procedure for extracting discriminant features, termed recursive cluster-based linear discriminant (RCLD), is proposed in this paper. Compared to the traditional Fisher linear discriminant (FLD) and its variations, RCLD has a number of advantages. First of all, it relaxes the constraint on the total number of features that can be extracted. Second, it fully exploits all information available for discrimination. In addition, RCLD is able to cope with multimodal distributions, which overcomes an inherent problem of conventional FLDs, which assumes uni-modal class distributions. Extensive experiments have been carried out on various types of face recognition problems for Yale, Olivetti Research Laboratory, and JAFFE databases to evaluate and compare the performance of the proposed algorithm with other feature extraction methods. The resulting improvement of performances by the new feature extraction scheme is significant.

Altered expression of β-catenin and UDP-glucose dehydrogenase in normal appearing acini of cancer-bearing prostate glands, detected by quantitative fluorescence imaging analysis

14540

Background: Expression of β-catenin and UDP-glucose dehydrogenase (UGDH) is markedly altered in prostate adenocarcinomas. The objective of this work was to evaluate expression of these proteins in normal appearing acini (NAA) of cancer-bearing prostate glands, as potential biomarkers of PC field disease. Methods: Archived core biopsies of 42 pairs of PC cases and controls were labeled for β-Catenin, and a tissue microarray of 10 prostatectomy specimens of cancer cases and controls were labeled for UGDH. A fluorophore-conjugated secondary antibody was incorporated into the staining protocol. Fluorescence in prostatic acini was measured with an automated Leica microscope and Image Pro Plus software and expressed as mean pixel intensity (MPI) per acinus and average MPI (AMPI) per case or control. Results: Mean AMPI of β-catenin in NAA of 33 cases (545 ± 93) was lower (p < 0.01) than that of normal acini (NA) in age-matched controls (684 ± 90), whereas mean AMPI of NAA in the remaining 9 cases (712 ± 111) was higher (p = 0.003) than that of the matched controls (634 ± 95). Twelve cases under-expressing β-catenin in NAA had AMPIs below the lowest AMPI (506 ± 83) of all controls. Mean AMPI of UGDH in NAA of 7 cases (372 ± 169) was lower (p < 0.05) in comparison to both NA (558 ± 119) and cancerous acini (CA) (645 ± 88). The difference between CA and NA did not achieve statistical significance. Conclusions: Reduced expression of β-catenin is a cancer-associated event in NAA. A cut-off AMPI (506) for β-catenin detected 29% of PC cases with 100% specificity. Altered expression of UGDH in NAA is a potential biomarker of PC. Funded by DoD grant DAMD17–02–1-0121.

No significant financial relationships to disclose.

Enhancing density-based data reduction using entropy

Data reduction algorithms determine a small data subset from a given large data set. In this article, new types of data reduction criteria, based on the concept of entropy, are first presented. These criteria can evaluate the data reduction performance in a sophisticated and comprehensive way. As a result, new data reduction procedures are developed. Using the newly introduced criteria, the proposed data reduction scheme is shown to be efficient and effective. In addition, an outlier-filtering strategy, which is computationally insignificant, is developed. In some instances, this strategy can substantially improve the performance of supervised data analysis. The proposed procedures are compared with related techniques in two types of application: density estimation and classification. Extensive comparative results are included to corroborate the contributions of the proposed algorithms.

A novel adenoviral vector which mediates hypoxia-inducible gene expression selectively in neurons

Selective gene expression in neurons is still a challenge. We have developed several expression vectors using a combination of neuron restrictive silencer elements (NRSEs), hypoxia responsive elements (HREs) and CMV minimal promoter (CMVmp). These elements were packaged into replication defective adenovirus to target gene expression selectively in neurons in a hypoxia-regulated manner. Neuronal selectivity and responsiveness to hypoxia of these novel constructs were determined empirically in both neural cell lines and primary cerebellar granule neurons (CGNs). The construct p5HRE-3NRSE exhibited not only the highest level of reporter gene expression in neuronal cells but also in an oxygen concentration-dependent manner when compared with all other constructs. As expected, this construct did not elicit reporter gene expression in non-neuronal cells including human HEK293A and HT29 cells, rat NRK cells, mouse 3T6 cells and 3T3 L1 cells. This construct was packaged into a replication defective adenoviral vector (Ad/5HRE-3NRSE) to determine neuron-selective and hypoxia-inducible gene expression in cultured mouse postmitotic primary CGNs and differentiated human NT2 neurons (NT2/Ns). Remarkably, in response to hypoxia, Ad/5HRE-3NRSE showed strong hypoxia-inducible gene expression selectively in neurons (12-fold induction in CGNs and 22-fold in NT2/Ns), but not in glial cells. Taken together, this vector with restricted gene expression to neurons under the regulation of hypoxia will be a useful tool for investigations of mechanisms of neuronal damage caused by ischemic insult.