Receiver operating characteristic analysis: a general tool for DNA array data filtration and performance estimation

A high-throughput integrated microfluidics method enables tyrosine autophosphorylation discovery

Autophosphorylation of receptor and non-receptor tyrosine kinases is a common molecular switch with broad implications for pathogeneses and therapy of cancer and other human diseases. Technologies for large-scale discovery and analysis of autophosphorylation are limited by the inherent difficulty to distinguish between phosphorylation and autophosphorylation in vivo and by the complexity associated with functional assays of receptors kinases in vitro. Here, we report a method for the direct detection and analysis of tyrosine autophosphorylation using integrated microfluidics and freshly synthesized protein arrays. We demonstrate the efficacy of our platform in detecting autophosphorylation activity of soluble and transmembrane tyrosine kinases, and the dependency of in vitro autophosphorylation assays on membranes. Our method, Integrated Microfluidics for Autophosphorylation Discovery (IMAD), is high-throughput, requires low reaction volumes and can be applied in basic and translational research settings. To our knowledge, it is the first demonstration of posttranslational modification analysis of membrane protein arrays.

Multichannel lung sound analysis for asthma detection

BACKGROUND AND OBJECTIVE

Lung sound signals convey valuable information of the lung status. Auscultation is an effective technique to appreciate the condition of the respiratory system using lung sound signals. The prior works on asthma detection from lung sound signals rely on the presence of wheeze. In this paper, we have classified normal and asthmatic subjects using advanced signal processing of posterior lung sound signals, even in the absence of wheeze.

METHODS

We collected lung sounds of 60 subjects (30 normal and 30 asthma) using a novel 4-channel data acquisition system from four different positions over the posterior chest, as suggested by the pulmonologist. A spectral subband based feature extraction scheme is proposed that works with artificial neural network (ANN) and support vector machine (SVM) classifiers for the multichannel signal. The power spectral density (PSD) is estimated from extracted lung sound cycle using Welch's method, which then decomposed into uniform subbands. A set of statistical features is computed from each subband and applied to ANN and SVM classifiers to classify normal and asthmatic subjects.

RESULTS

In the first part of this study, the performances of each individual channel and four channels together are evaluated where the combined channel performance is found superior to that of individual channels. Next, the performances of all possible combinations of the channels are investigated and the best classification accuracies of 89.2( ± 3.87)% and 93.3( ± 3.10)% are achieved for 2-channel and 3-channel combinations in ANN and SVM classifiers, respectively.

CONCLUSIONS

The proposed multichannel asthma detection method where the presence of wheeze in lung sound is not a necessary requirement, outperforms commonly used lung sound classification methods in this field and provides significant relative improvement. The channel combination study gives insight into the contribution of respective lung sound collection areas and their combinations in asthma detection.

Gene selection for microarray cancer classification using a new evolutionary method employing artificial intelligence concepts

Gene selection is a demanding task for microarray data analysis. The diverse complexity of different cancers makes this issue still challenging. In this study, a novel evolutionary method based on genetic algorithms and artificial intelligence is proposed to identify predictive genes for cancer classification. A filter method was first applied to reduce the dimensionality of feature space followed by employing an integer-coded genetic algorithm with dynamic-length genotype, intelligent parameter settings, and modified operators. The algorithmic behaviors including convergence trends, mutation and crossover rate changes, and running time were studied, conceptually discussed, and shown to be coherent with literature findings. Two well-known filter methods, Laplacian and Fisher score, were examined considering similarities, the quality of selected genes, and their influences on the evolutionary approach. Several statistical tests concerning choice of classifier, choice of dataset, and choice of filter method were performed, and they revealed some significant differences between the performance of different classifiers and filter methods over datasets. The proposed method was benchmarked upon five popular high-dimensional cancer datasets; for each, top explored genes were reported. Comparing the experimental results with several state-of-the-art methods revealed that the proposed method outperforms previous methods in DLBCL dataset.

A comprehensive and universal method for assessing the performance of differential gene expression analyses

The number of methods for pre-processing and analysis of gene expression data continues to increase, often making it difficult to select the most appropriate approach. We present a simple procedure for comparative estimation of a variety of methods for microarray data pre-processing and analysis. Our approach is based on the use of real microarray data in which controlled fold changes are introduced into 20% of the data to provide a metric for comparison with the unmodified data. The data modifications can be easily applied to raw data measured with any technological platform and retains all the complex structures and statistical characteristics of the real-world data. The power of the method is illustrated by its application to the quantitative comparison of different methods of normalization and analysis of microarray data. Our results demonstrate that the method of controlled modifications of real experimental data provides a simple tool for assessing the performance of data preprocessing and analysis methods.

MUC1-associated proliferation signature predicts outcomes in lung adenocarcinoma patients

Background

MUC1 protein is highly expressed in lung cancer. The cytoplasmic domain of MUC1 (MUC1-CD) induces tumorigenesis and resistance to DNA-damaging agents. We characterized MUC1-CD-induced transcriptional changes and examined their significance in lung cancer patients.

Methods

Using DNA microarrays, we identified 254 genes that were differentially expressed in cell lines transformed by MUC1-CD compared to control cell lines. We then examined expression of these genes in 441 lung adenocarcinomas from a publicly available database. We employed statistical analyses independent of clinical outcomes, including hierarchical clustering, Student's t-tests and receiver operating characteristic (ROC) analysis, to select a seven-gene MUC1-associated proliferation signature (MAPS). We demonstrated the prognostic value of MAPS in this database using Kaplan-Meier survival analysis, log-rank tests and Cox models. The MAPS was further validated for prognostic significance in 84 lung adenocarcinoma patients from an independent database.

Results

MAPS genes were found to be associated with proliferation and cell cycle regulation and included CCNB1, CDC2, CDC20, CDKN3, MAD2L1, PRC1 and RRM2. MAPS expressors (MAPS+) had inferior survival compared to non-expressors (MAPS-). In the initial data set, 5-year survival was 65% (MAPS-) vs. 45% (MAPS+, p < 0.0001). Similarly, in the validation data set, 5-year survival was 57% (MAPS-) vs. 28% (MAPS+, p = 0.005).

Conclusions

The MAPS signature, comprised of MUC1-CD-dependent genes involved in the control of cell cycle and proliferation, is associated with poor outcomes in patients with adenocarcinoma of the lung. These data provide potential new prognostic biomarkers and treatment targets for lung adenocarcinoma.

Ad.Egr-TNF and local ionizing radiation suppress metastases by interferon-beta-dependent activation of antigen-specific CD8+ T cells

Ad.Egr-TNF is a radioinducible adenovector currently in phase 3 trials for inoperable pancreatic cancer. The combination of Ad.Egr-TNF and ionizing radiation (IR) contributes to local tumor control through the production of tumor necrosis factor-alpha (TNFalpha) in the tumor microenvironment. Moreover, clinical and preclinical studies with Ad.Egr-TNF/IR have suggested that this local approach suppresses the growth of distant metastatic disease; however, the mechanisms responsible for this effect remain unclear. These studies have been performed in wild-type (WT) and TNFR1,2(-/-) mice to assess the role of TNFalpha-induced signaling in the suppression of draining lymph node (DLN) metastases. The results demonstrate that production of TNFalpha in the tumor microenvironment induces expression of interferon (IFNbeta). In turn, IFNbeta stimulates the production of chemokines that recruit CD8(+) T cells to the tumor. The results further demonstrate that activation of tumor antigen-specific CD8(+) CTLs contributes to local antitumor activity and suppression of DLN metastases. These findings support a model in which treatment of tumors with Ad.Egr-TNF and IR is mediated by local and distant immune-mediated antitumor effects that suppress the development of metastases.

Cooperativity of the MUC1 oncoprotein and STAT1 pathway in poor prognosis human breast cancer

Signal transducer and activator of transcription 1 (STAT1) is activated in the inflammatory response to interferons. The MUC1 oncoprotein is overexpressed in human breast cancers. Analysis of genes differentially expressed in MUC1-transformed cells has identified a network linking MUC1 and STAT1 that is associated with cellular growth and inflammation. The results further demonstrate that the MUC1-C subunit associates with STAT1 in cells and that the MUC1-C cytoplasmic domain binds directly to the STAT1 DNA binding domain. The interaction between MUC1-C and STAT1 is inducible by IFNγ in non-malignant epithelial cells and constitutive in breast cancer cells. Moreover, the MUC1-STAT1 interaction contributes to the activation of STAT1 target genes, including MUC1 itself. Analysis of two independent databases demonstrated that MUC1 and STAT1 are coexpressed in about 15% of primary human breast tumors. Coexpression of MUC1 and the STAT1 pathway was found to be significantly associated with decreased recurrence-free and overall survival. These findings indicate that (i) MUC1 and STAT1 function in an auto-inductive loop, and (ii) activation of both MUC1 and the STAT1 pathway in breast tumors confers a poor prognosis for patients.

STAT1-dependent expression of energy metabolic pathways links tumour growth and radioresistance to the Warburg effect

BACKGROUND

The Signal Transducer and Activator of Transcription 1 (STAT1) has traditionally been regarded as a transmitter of interferon signaling and a pro-apoptotic tumour suppressor. Recent data have identified new functions of STAT1 associated with tumourigenesis and resistance to genotoxic stress, including ionizing radiation (IR) and chemotherapy. To investigate the mechanisms contributing to the tumourigenic functions of STAT1, we performed a combined transcriptomic-proteomic expressional analysis and found that STAT1 is associated with regulation of energy metabolism with potential implication in the Warburg effect.

METHODS

We generated a stable knockdown of STAT1 in the SCC61 human squamous cell carcinoma cell line, established tumour xenografts in athymic mice, and compared transcriptomic and proteomic profiles of STAT1 wild-type (WT) and knockdown (KD) untreated or irradiated (IR) tumours. Transcriptional profiling was based on Affymetrix Human GeneChip(R) Gene 1.0 ST microarrays. Proteomes were determined from the tandem mass spectrometry (MS/MS) data by searching against the human subset of the UniProt database. Data were analysed using Significance Analysis of Microarrays for ribonucleic acid and Visualize software for proteins. Functional analysis was performed with Ingenuity Pathway Analysis with statistical significance measured by Fisher's exact test.

RESULTS

Knockdown of STAT1 led to significant growth suppression in untreated tumours and radio sensitization of irradiated tumours. These changes were accompanied by alterations in the expression of genes and proteins of glycolysis/gluconeogenesis (GG), the citrate cycle (CC) and oxidative phosphorylation (OP). Of these pathways, GG had the most concordant changes in gene and protein expression and demonstrated a STAT1-dependent expression of genes and proteins consistent with tumour-specific glycolysis. In addition, IR drastically suppressed the GG pathway in STAT1 KD tumours without significant change in STAT1 WT tumours.

CONCLUSION

Our results identify a previously uncharacterized function of STAT1 in tumours: expressional regulation of genes encoding proteins involved in glycolysis, the citrate cycle and mitochondrial oxidative phosphorylation, with predominant regulation of glycolytic genes. STAT1-dependent expressional regulation of glycolysis suggests a potential role for STAT1 as a transcriptional modulator of genes responsible for the Warburg effect.

Internal standard-based analysis of microarray data. Part 1: analysis of differential gene expressions

Genome-scale microarray experiments for comparative analysis of gene expressions produce massive amounts of information. Traditional statistical approaches fail to achieve the required accuracy in sensitivity and specificity of the analysis. Since the problem can be resolved neither by increasing the number of replicates nor by manipulating thresholds, one needs a novel approach to the analysis. This article describes methods to improve the power of microarray analyses by defining internal standards to characterize features of the biological system being studied and the technological processes underlying the microarray experiments. Applying these methods, internal standards are identified and then the obtained parameters are used to define (i) genes that are distinct in their expression from background; (ii) genes that are differentially expressed; and finally (iii) genes that have similar dynamical behavior.

Pattern recognition methods applied to respiratory sounds classification into normal and wheeze classes

In this paper, we present the pattern recognition methods proposed to classify respiratory sounds into normal and wheeze classes. We evaluate and compare the feature extraction techniques based on Fourier transform, linear predictive coding, wavelet transform and Mel-frequency cepstral coefficients (MFCC) in combination with the classification methods based on vector quantization, Gaussian mixture models (GMM) and artificial neural networks, using receiver operating characteristic curves. We propose the use of an optimized threshold to discriminate the wheezing class from the normal one. Also, post-processing filter is employed to considerably improve the classification accuracy. Experimental results show that our approach based on MFCC coefficients combined to GMM is well adapted to classify respiratory sounds in normal and wheeze classes. McNemar's test demonstrated significant difference between results obtained by the presented classifiers (p<0.05).

MUC1-induced transcriptional programs associated with tumorigenesis predict outcome in breast and lung cancer

The Mucin 1 (MUC1) oncoprotein is aberrantly overexpressed in diverse human malignancies including breast and lung cancer. Although MUC1 modulates the activity of several transcription factors, there is no information regarding the effects of MUC1 on global gene expression patterns and the potential role of MUC1-induced genes in predicting outcome for cancer patients. We have developed an experimental model of MUC1-induced transformation that has identified the activation of gene families involved in oncogenesis, angiogenesis, and extracellular matrix remodeling. A set of experimentally derived MUC1-induced genes associated with tumorigenesis was applied to the analysis of breast and lung adenocarcinoma cancer databases. A 35-gene MUC1-induced tumorigenesis signature predicts significant decreases in both disease-free and overall survival in patients with breast (n=295) and lung (n=442) cancers. The data show that the MUC1 oncoprotein contributes to the regulation of genes that are highly predictive of clinical outcome in breast and lung cancer patients.

MUC1-induced alterations in a lipid metabolic gene network predict response of human breast cancers to tamoxifen treatment

The mucin 1 (MUC1) oncoprotein is aberrantly overexpressed in human breast cancers. Although MUC1 modulates the activity of estrogen receptor alpha (ER), there is no information regarding the effects of MUC1 on global gene expression patterns and the potential role of MUC1-induced genes in predicting outcome for breast cancer patients. We have developed an experimental model of MUC1-induced transformation that has identified the activation of genes involved in cholesterol and fatty acid metabolism. A 38-gene set of experimentally derived MUC1-induced genes associated with lipid metabolism was applied to the analysis of ER(+) breast cancer patients treated with tamoxifen. The results obtained from 2 independent databases demonstrate that patients overexpressing MUC1 and the lipid metabolic pathways are at significantly higher risk for death and recurrence/distant metastasis. By contrast, these genes were not predictive in untreated patients. Furthermore, a positive correlation was found between expression of the 38-gene set and the ER signaling pathway. These findings indicate that (i) MUC1 regulates cholesterol and fatty acid metabolism, and (ii) activation of these pathways in ER(+) breast cancers predicts failure to tamoxifen treatment.

Signal deconvolution based expression-detection and background adjustment for microarray data

Background adjustment is an essential stage in analyzing DNA microarrays. Discriminating expressed genes from unexpressed ones (expression detection), and estimating the expression levels of weakly expressed genes, critically depend on accurate treatment of the background intensity. Current methods for background adjustment either do not deal with nonspecific hybridization or strongly depend on the reliability of control probes. Existing model-based methods have limited accuracy. A new platform-independent background adjustment algorithm is presented. The algorithm relies on the deconvoluted experimental signal distribution for evaluating the expression probability and adjusting the background of each probe. Considering expression detection, it is shown, for two-channels cDNA arrays and for the Affymetrix GeneChip platform, that the algorithm performs at least as good or better than control-probes-based algorithms. For the Affymetrix GeneChip arrays, it is further shown that the algorithm outperforms the robust multiarray (RMA) expression measure in estimating genomewide expression levels.

Reagent preparation and storage for amplification of microarray hybridization targets with a fully automated system

The advent of automated systems for gene expression profiling has accentuated the need for the development of convenient and cost-effective methods for reagent preparation. We have developed a method for the preparation and storage of pre-aliquoted cocktail plates that contain all reagents required for amplification of nucleic acid by reverse transcription and in vitro transcription reactions. Plates can be stored at -80 degrees C for at least 1 month and kept in a hotel at 4 degrees C for at least 24 h prior to use. Microarray data quality generated from these pre-aliquoted reagent plates is not statistically different between cRNA amplified with stored cocktails and cRNA amplified with freshly prepared cocktails. Deployment of pre-aliquoted, stored cocktail plates in a fully automated system not only increases the throughput of amplifying cRNA targets from thousands of RNA samples, but could also considerably reduce reagent costs and potentially improve process robustness.

Tentacle probes: eliminating false positives without sacrificing sensitivity

The majority of efforts to increase specificity or sensitivity in biosensors result in trade-offs with little to no gain in overall accuracy. This is because a biosensor cannot be more accurate than the affinity interaction it is based on. Accordingly, we have developed a new class of reagents based on mathematical principles of cooperativity to enhance the accuracy of the affinity interaction. Tentacle probes (TPs) have a hairpin structure similar to molecular beacons (MBs) for enhanced specificity, but are modified by the addition of a capture probe for increased kinetics and affinity. They produce kinetic rate constants up to 200-fold faster than MB with corresponding stem strengths. Concentration-independent specificity was observed with no false positives at up to 1 mM concentrations of variant analyte. In contrast, MBs were concentration dependent and experienced false positives above 3.88 μM of variant analyte. The fast kinetics of this label-free reagent may prove important for extraction efficiency, hence sensitivity and detection time, in microfluidic assays. The concentration-independent specificity of TPs may prove extremely useful in assays where starting concentrations and purities are unknown as would be the case in bioterror or clinical point of care diagnostics.

Progression of Barrett's metaplasia to adenocarcinoma is associated with the suppression of the transcriptional programs of epidermal differentiation

We did expressional profiling on 24 paired samples of normal esophageal epithelium, Barrett's metaplasia, and esophageal adenocarcinomas. Matching tissue samples representing the three different histologic types were obtained from each patient undergoing esophagectomy for adenocarcinoma. Our analysis compared the molecular changes accompanying the transformation of normal squamous epithelium with Barrett's esophagus and adenocarcinoma in individual patients rather than in a random cohort. We tested the hypothesis that expressional profiling may reveal gene sets that can be used as molecular markers of progression from normal esophageal epithelium to Barrett's esophagus and adenocarcinoma. Expressional profiling was done using U133A GeneChip (Affymetrix), which represent approximately two thirds of the human genome. The final selection of 214 genes permitted the discrimination of differential gene expression of normal esophageal squamous epithelium, Barrett's esophagus, and adenocarcinoma using two-dimensional hierarchical clustering of selected genes. These data indicate that transformation of Barrett's esophagus to adenocarcinoma is associated with suppression of the genes involved in epidermal differentiation, including genes in 1q21 loci and corresponding to the epidermal differentiation complex. Correlation analysis of genes concordantly expressed in Barrett's esophagus and adenocarcinoma revealed 21 genes that represent potential genetic markers of disease progression and pharmacologic targets for treatment intervention. PCR analysis of genes selected based on DNA array experiments revealed that estimation of the ratios of GATA6 to SPRR3 allows discrimination among normal esophageal epithelium, Barrett's dysplasia, and adenocarcinoma.

TAp63γ can substitute for p53 in inducing expression of themaspintumor suppressor

Maspin is a Class II tumor suppressor protein and plays a role in tumor growth by inhibiting cellular invasion and motility. It is a member of the serpin family of protease inhibitors and has been shown to reduce angiogenesis. Maspin gene expression can be upregulated by the tumor suppressor p53. We tested 7 p53-related proteins of the p63 and p73 families for their ability to induce maspin expression. The p63 splice form TAp63gamma can substitute for p53 in activating the maspin promoter. TAp63gamma activates the promoter through the same consensus site as p53. In the DLD-1 colorectal adenocarcinoma cell line, harboring a tet-off regulated transgene, induction of TAp63gamma leads to an upregulation of maspin mRNA from the chromosomal gene. With a short lag phase also maspin protein levels are elevated after induced TAp63gamma expression. To assess a potential function of p63-dependent maspin upregulation in tumors we followed expression of p53, p63 and maspin by immunohistochemistry in hepatocellular carcinomas. Two types of tumors with wild-type or mutant p53 were assayed. Interestingly, the majority of tumors expressing only a mutated and inactive p53 protein nonetheless stain positive for maspin, whereas these tumors were positive for p63 protein expression. In summary, we show that TAp63gamma can substitute for p53 in transcriptional activation of the maspin tumor suppressor gene. TAp63gamma employs the same DNA recognition site for this activation as p53. We observe expression patterns of p53, p63 and maspin proteins in tumor tissue that may indicate also a function of maspin induction by p63 in tumors.

Interaction of amifostine and ionizing radiation on transcriptional patterns of apoptotic genes expressed in human microvascular endothelial cells (HMEC)

PURPOSE

Amifostine is a prodrug that requires dephosphorylation by alkaline phosphatase to become activated. This process occurs rapidly within the bloodstream after its i.v. administration to patients undergoing cancer treatment with selected radiation and chemotherapies. Vascular endothelial cells will, therefore, represent a normal cell system that is among the first to experience the radioprotective effects of this agent. Amifostine's active free thiol WR-1065 was investigated to determine its effect on radiation-induced changes in transcriptional patterns and subsequent apoptosis in human microvascular endothelial cells (HMEC) growing in vitro.

METHODS AND MATERIALS

Human microvascular endothelial cells were grown to confluency and then exposed to WR-1065 at a concentration of 4 mM for 30 min, radiation doses that ranged from 0 to 6 Gy, and WR-1065 at a concentration of 4 mM for 30 min before exposure to ionizing radiation. Cell survival was assessed by clonogenic assay, cell cycle phase was analyzed by flow cytometry, apoptosis was also assessed by flow cytometry in which Anexin V staining and sub-G1 fraction analysis were applied, and gene expression was analyzed by the Clontech Atlas Human cDNA array to identify synergistic and antagonistic effects as a function of amifostine and radiation exposure conditions with a focus on apoptotic-related factors.

RESULTS

Exposure of HMEC to 4 mM WR-1065 30 min before irradiation resulted in a protection enhancement factor of 2.0; that is, D(O-IRR) of 1.25 Gy and D(O-IRR+WR) of 2.56 Gy. Expression profiling revealed 29 genes that were synergistically activated by the combined action of WR-1065 and ionizing radiation, and an additional 12 genes were synergistically or additively suppressed. In particular, a subset of apoptosis-related genes that included caspases 2, 4, and 9 and different members of the bcl family, along with apoptosis-related receptors, were identified as being significantly affected by the combined treatment of WR-1065 and radiation exposure. In addition, a number of cell cycle-related genes that express cyclins A, G1, G2, and D3 and DNA damage/check point proteins ATM, DNA-PK and RAD23B were also found to be significantly affected. Functional assays of apoptosis were also performed that demonstrated the ability of WR-1065 to protect against radiation-induced apoptosis.

CONCLUSIONS

WR-1065, the active thiol form of amifostine, is an effective radioprotector of HMEC as determined by use of clonogenic and apoptotic assays for cell survival. Expression profiling successfully defined the transcriptional response of HMEC to both WR-1065 and ionizing radiation exposure, either alone or in combination, and demonstrated both synergistic and antagonistic effects on the expression of different cellular genes, along with corresponding functional responses. The radioprotective effects of amifostine are not limited to its well-characterized physiochemical properties, which include free-radical scavenging, auto-oxidation leading to intracellular hypoxia, and chemical repair by hydrogen atom donation, but include its ability to modulate the complex transcriptional regulation of genes that are involved in apoptosis, cell cycle, and DNA repair.

Rank Difference Analysis of Microarrays (RDAM), a novel approach to statistical analysis of microarray expression profiling data

Background

A key step in the analysis of microarray expression profiling data is the identification of genes that display statistically significant changes in expression signals between two biological conditions.

Results

We describe a new method, Rank Difference Analysis of Microarrays (RDAM), which estimates the total number of truly varying genes and assigns a p-value to each signal variation. Information on a group of differentially expressed genes includes the sensitivity and the false discovery rate. We demonstrate the feasibility and efficiency of our approach by applying it to a large synthetic expression data set and to a biological data set obtained by comparing vegetatively-growing wild type and tor2-mutant yeast strains. In both cases we observed a significant improvement of the power of analysis when our method is compared to another popular nonparametric method.

Conclusions

This study provided a valuable new statistical method to analyze microarray data. We conclude that the good quality of the results obtained by RDAM is mainly due to the quasi-perfect equalization of variation distribution, which is related to the standardization procedure used and to the measurement of variation by rank difference.

Prediction of splice sites with dependency graphs and their expanded bayesian networks

MOTIVATION

Owing to the complete sequencing of human and many other genomes, huge amounts of DNA sequence data have been accumulated. In bioinformatics, an important issue is how to predict the complete structure of genes from the genomic DNA sequence, especially the human genome. A crucial part in the gene structure prediction is to determine the precise exon-intron boundaries, i.e. the splice sites, in the coding region.

RESULTS

We have developed a dependency graph model to fully capture the intrinsic interdependency between base positions in a splice site. The establishment of dependency between two position is based on a chi2-test from known sample data. To facilitate statistical inference, we have expanded the dependency graph (which is usually a graph with cycles that make probabilistic reasoning very difficult, if not impossible) into a Bayesian network (which is a directed acyclic graph that facilitates statistical reasoning). When compared with the existing models such as weight matrix model, weight array model, maximal dependence decomposition, Cai et al.'s tree model as well as the less-studied second-order and third-order Markov chain models, the expanded Bayesian networks from our dependency graph models perform the best in nearly all the cases studied.

AVAILABILITY

Software (a program called DGSplicer) and datasets used are available at http://csrl.ee.nthu.edu.tw/bioinf/

CONTACT

cclu@ee.nthu.edu.tw.

Endothelial cells co-cultured with wild-type and dominant/negative p53-transfected glioblastoma cells exhibit differential sensitivity to radiation-induced apoptosis

We performed expressional profiling of isogenic glioblastoma cell lines U87-Lux8 and U87-175.4. These cell lines differ in that U87-Lux8 expresses wild-type p53 and U87-175.4 expresses a dominant-negative p53 (175(His) mutation). DNA array analysis and real-time PCR measurements demonstrated that basal expression and response to irradiation were different in these isogenic glioblastoma cell lines. These differences included genes involved in growth regulation and genes associated with cell-to-cell and cell/ECM communications. Co-cultivation of U87-175.4 and U87-Lux8 with HUVE cells demonstrated that U87-175.4 cells suppress the angiogenic phenotype of HUVEC and increase their sensitivity to radiation-induced apoptosis compared to co-culture of U87-Lux8/HUVEC. These data suggest that blockade of p53 function may alter the communication between tumor cells and endothelial cells such that endothelial cells exhibit an increase in radiosensitivity. These findings may have important implications for the treatment of glioblastoma tumors and other human cancers.

A microarray platform comparison for neuroscience applications

To address the need for high sensitivity in gene expression profiling of small neural tissue samples ( approximately 100 ng total RNA), we compared a novel RT-PCR-IVT protocol using fluor-reverse pairs on inkjet oligonucleotide microarrays and an RT-IVT protocol using 33P labeling on nylon cDNA arrays. The comparison protocol was designed to evaluate these systems for sensitivity, specificity, reproducibility, and linearity. We developed parameters, thresholds, and testing conditions that could be used to differentiate various systems that spanned detection chemistry and instrumentation; probe number and selection criteria; and sample processing protocols. We concluded that the inkjet system had better performance in sensitivity, specificity, and reproducibility than the nylon system, and similar performance in linearity. Between these two platforms, the data indicates that the inkjet system would perform better for the transcriptional profiling of 100 ng total RNA samples for neuroscience studies.

STAT1 is overexpressed in tumors selected for radioresistance and confers protection from radiation in transduced sensitive cells

Nu61, a radiation-resistant human tumor xenograft, was selected from a parental radiosensitive tumor SCC-61 by eight serial cycles of passage in athymic nude mice and in vivo irradiation. Replicate DNA array experiments identified 52 genes differentially expressed in nu61 tumors compared with SCC-61 tumors. Of these, 19 genes were in the IFN-signaling pathway and moreover, 25 of the 52 genes were inducible by IFN in the nu61 cell line. Among the genes involved in IFN signaling, STAT1alpha and STAT1beta were the most highly overexpressed in nu61 compared to SCC-61. STAT1alpha and STAT1beta cDNAs were cloned and stably transfected into SCC-61 tumor cells. Clones of SCC-61 tumor cells transfected with vectors expressing STAT1alpha and STAT1beta demonstrated radioprotection after exposure to 3 Gy (P < 0.038). The results indicate that radioresistance acquired during radiotherapy treatment may account for some treatment failures and demonstrate an association of acquired tumor radioresistance with up-regulation of components of the IFN-related signaling pathway.