Machine learning approaches to decipher hormone and HER2 receptor status phenotypes in breast cancer.

Breast cancer prognosis and administration of therapies are aided by knowledge of hormonal and HER2 receptor status. Breast cancer lacking estrogen receptors, progesterone receptors and HER2 receptors are difficult to treat. Regarding large data repositories such as The Cancer Genome Atlas, available wet-lab methods for establishing the presence of these receptors do not always conclusively cover all available samples. To this end, we introduce median-supplement methods to identify hormonal and HER2 receptor status phenotypes of breast cancer patients using gene expression profiles. In these approaches, supplementary instances based on median patient gene expression are introduced to balance a training set from which we build simple models to identify the receptor expression status of patients. In addition, for the purpose of benchmarking, we examine major machine learning approaches that are also applicable to the problem of finding receptor status in breast cancer. We show that our methods are robust and have high sensitivity with extremely low false-positive rates compared with the well-established methods. A successful application of these methods will permit the simultaneous study of large collections of samples of breast cancer patients as well as save time and cost while standardizing interpretation of outcomes of such studies.

High-fat diet induces fibrosis in mice lacking CYP2A5 and PPARα: a new model for steatohepatitis-associated fibrosis.

Obesity is linked to nonalcoholic steatohepatitis. Peroxisome proliferator-activated receptor-α (PPARα) regulates lipid metabolism. Cytochrome P-450 2A5 (CYP2A5) is a potential antioxidant and CYP2A5 induction by ethanol is CYP2E1 dependent. High-fat diet (HFD)-induced obesity and steatosis are more severe in CYP2A5 knockout (cyp2a5-/-) mice than in wild-type mice although PPARα is elevated in cyp2a5-/- mice. To examine why the upregulated PPARα failed to prevent the enhanced steatosis in cyp2a5-/- mice, we abrogate the upregulated PPARα in cyp2a5-/- mice by cross-breeding cyp2a5-/- mice with PPARα knockout (pparα-/-) mice to create pparα-/-/cyp2a5-/- mice. The pparα-/-/cyp2a5-/- mice, pparα-/- mice, and cyp2a5-/- mice were fed HFD to induce steatosis. After HFD feeding, more severe steatosis was developed in pparα-/-/cyp2a5-/- mice than in pparα-/- mice and cyp2a5-/- mice. The pparα-/-/cyp2a5-/- mice and pparα-/- mice exhibited comparable and impaired lipid metabolism. Elevated serum alanine transaminase and liver interleukin-1ß, liver inflammatory cell infiltration, and foci of hepatocellular ballooning were observed in pparα-/-/cyp2a5-/- mice but not in pparα-/- mice and cyp2a5-/- mice. In pparα-/-/cyp2a5-/- mice, although redox-sensitive transcription factor nuclear factor erythroid 2-related factor 2 and its target antioxidant genes were upregulated as a compensation, thioredoxin was suppressed, and phosphorylation of JNK and formation of nitrotyrosine adduct were increased. Liver glutathione was decreased, and lipid peroxidation was increased. Interestingly, inflammation and fibrosis were all observed within the clusters of lipid droplets, and these lipid droplet clusters were all located inside the area with CYP2E1-positive staining. These results suggest that HFD-induced fibrosis in pparα-/-/cyp2a5-/- mice is associated with steatosis, and CYP2A5 interacts with PPARα to participate in regulating steatohepatitis-associated fibrosis.

De novo lipogenesis represents a therapeutic target in mutant Kras non-small cell lung cancer.

Oncogenic Kras mutations are one of the most common alterations in non-small cell lung cancer and are associated with poor response to treatment and reduced survival. Driver oncogenes, such as Kras are now appreciated for their ability to promote tumor growth via up-regulation of anabolic pathways. Therefore, we wanted to identify metabolic vulnerabilities in Kras-mutant lung cancer. Using the Kras LSL-G12D lung cancer model, we show that mutant Kras drives a lipogenic gene-expression program. Stable-isotope analysis reveals that mutant Kras promotes de novo fatty acid synthesis in vitro and in vivo. The importance of fatty acid synthesis in Kras-induced tumorigenesis was evident by decreased tumor formation in Kras LSL-G12D mice after treatment with a fatty acid synthesis inhibitor. Importantly, with gain and loss of function models of mutant Kras, we demonstrate that mutant Kras potentiates the growth inhibitory effects of several fatty acid synthesis inhibitors. These studies highlight the potential to target mutant Kras tumors by taking advantage of the lipogenic phenotype induced by mutant Kras.-Singh, A., Ruiz, C., Bhalla, K., Haley, J. A., Li, Q. K., Acquaah-Mensah, G., Montal, E., Sudini, K. R., Skoulidis, F., Wistuba, I. I., Papadimitrakopoulou, V., Heymach, J. V., Boros, L. G., Gabrielson, E., Carretero, J., Wong, K.-k., Haley, J. D., Biswal, S., Girnun, G. D. De novo lipogenesis represents a therapeutic target in mutant Kras non-small cell lung cancer.

SAGA: a hybrid search algorithm for Bayesian Network structure learning of transcriptional regulatory networks.

Bayesian Networks have been used for the inference of transcriptional regulatory relationships among genes, and are valuable for obtaining biological insights. However, finding optimal Bayesian Network (BN) is NP-hard. Thus, heuristic approaches have sought to effectively solve this problem. In this work, we develop a hybrid search method combining Simulated Annealing with a Greedy Algorithm (SAGA). SAGA explores most of the search space by undergoing a two-phase search: first with a Simulated Annealing search and then with a Greedy search. Three sets of background-corrected and normalized microarray datasets were used to test the algorithm. BN structure learning was also conducted using the datasets, and other established search methods as implemented in BANJO (Bayesian Network Inference with Java Objects). The Bayesian Dirichlet Equivalence (BDe) metric was used to score the networks produced with SAGA. SAGA predicted transcriptional regulatory relationships among genes in networks that evaluated to higher BDe scores with high sensitivities and specificities. Thus, the proposed method competes well with existing search algorithms for Bayesian Network structure learning of transcriptional regulatory networks.

A Lung Cancer Mouse Model Database.

Lung cancer, the leading cause of cancer mortality, exhibits diverse histological subtypes and genetic complexities. Numerous preclinical mouse models have been developed to study lung cancer, but data from these models are disparate, siloed, and difficult to compare in a centralized fashion. Here we established the Lung Cancer Mouse Model Database (LCMMDB), an extensive repository of 1,354 samples from 77 transcriptomic datasets covering 974 samples from genetically engineered mouse models (GEMMs), 368 samples from carcinogen-induced models, and 12 samples from a spontaneous model. Meticulous curation and collaboration with data depositors have produced a robust and comprehensive database, enhancing the fidelity of the genetic landscape it depicts. The LCMMDB aligns 859 tumors from GEMMs with human lung cancer mutations, enabling comparative analysis and revealing a pressing need to broaden the diversity of genetic aberrations modeled in GEMMs. Accompanying this resource, we developed a web application that offers researchers intuitive tools for in-depth gene expression analysis. With standardized reprocessing of gene expression data, the LCMMDB serves as a powerful platform for cross-study comparison and lays the groundwork for future research, aiming to bridge the gap between mouse models and human lung cancer for improved translational relevance.

Suppressed expression of T-box transcription factors is involved in senescence in chronic obstructive pulmonary disease.

Chronic obstructive pulmonary disease (COPD) is a major global health problem. The etiology of COPD has been associated with apoptosis, oxidative stress, and inflammation. However, understanding of the molecular interactions that modulate COPD pathogenesis remains only partly resolved. We conducted an exploratory study on COPD etiology to identify the key molecular participants. We used information-theoretic algorithms including Context Likelihood of Relatedness (CLR), Algorithm for the Reconstruction of Accurate Cellular Networks (ARACNE), and Inferelator. We captured direct functional associations among genes, given a compendium of gene expression profiles of human lung epithelial cells. A set of genes differentially expressed in COPD, as reported in a previous study were superposed with the resulting transcriptional regulatory networks. After factoring in the properties of the networks, an established COPD susceptibility locus and domain-domain interactions involving protein products of genes in the generated networks, several molecular candidates were predicted to be involved in the etiology of COPD. These include COL4A3, CFLAR, GULP1, PDCD1, CASP10, PAX3, BOK, HSPD1, PITX2, and PML. Furthermore, T-box (TBX) genes and cyclin-dependent kinase inhibitor 2A (CDKN2A), which are in a direct transcriptional regulatory relationship, emerged as preeminent participants in the etiology of COPD by means of senescence. Contrary to observations in neoplasms, our study reveals that the expression of genes and proteins in the lung samples from patients with COPD indicate an increased tendency towards cellular senescence. The expression of the anti-senescence mediators TBX transcription factors, chromatin modifiers histone deacetylases, and sirtuins was suppressed; while the expression of TBX-regulated cellular senescence markers such as CDKN2A, CDKN1A, and CAV1 was elevated in the peripheral lung tissue samples from patients with COPD. The critical balance between senescence and anti-senescence factors is disrupted towards senescence in COPD lungs.

Multi-omics inference of differential breast cancer-related transcriptional regulatory network gene hubs between young Black and White patients.

OBJECTIVE: Breast cancers (BrCA) are a leading cause of illness and mortality worldwide. Black women have a higher incidence rate relative to white women prior to age 40 years, and a lower incidence rate after 50 years. The objective of this study is to identify -omics differences between the two breast cancer cohorts to better understand the disparities observed in patient outcomes. MATERIALS AND METHODS: Using Standard SQL, we queried ISB-CGC hosted Google BigQuery tables storing TCGA BrCA gene expression, methylation, and somatic mutation data and analyzed the combined multi-omics results using a variety of methods. RESULTS: Among Stage II patients 50 years or younger, genes PIK3CA and CDH1 are more frequently mutated in White (W50) than in Black or African American patients (BAA50), while HUWE1, HYDIN, and FBXW7 mutations are more frequent in BAA50. Over-representation analysis (ORA) and Gene Set Enrichment Analysis (GSEA) results indicate that, among others, the Reactome Signaling by ROBO Receptors gene set is enriched in BAA50. Using the Virtual Inference of Protein-activity by Enriched Regulon analysis (VIPER) algorithm, putative top 20 master regulators identified include NUPR1, NFKBIL1, ZBTB17, TEAD1, EP300, TRAF6, CACTIN, and MID2. CACTIN and MID2 are of prognostic value. We identified driver genes, such as OTUB1, with suppressed expression whose DNA methylation status were inversely correlated with gene expression. Networks capturing microRNA and gene expression correlations identified notable microRNA hubs, such as miR-93 and miR-92a-2, expressed at higher levels in BAA50 than in W50. DISCUSSION/CONCLUSION: The results point to several driver genes as being involved in the observed differences between the cohorts. The findings here form the basis for further mechanistic exploration.

A Tale of Two Diseases: Exploring Mechanisms Linking Diabetes Mellitus with Alzheimer's Disease.

Dementias, including the type associated with Alzheimer's disease (AD), are on the rise worldwide. Similarly, type 2 diabetes mellitus (T2DM) is one of the most prevalent chronic diseases globally. Although mechanisms and treatments are well-established for T2DM, there remains much to be discovered. Recent research efforts have further investigated factors involved in the etiology of AD. Previously perceived to be unrelated diseases, commonalities between T2DM and AD have more recently been observed. As a result, AD has been labeled as "type 3 diabetes". In this review, we detail the shared processes that contribute to these two diseases. Insulin resistance, the main component of the pathogenesis of T2DM, is also present in AD, causing impaired brain glucose metabolism, neurodegeneration, and cognitive impairment. Dysregulation of insulin receptors and components of the insulin signaling pathway, including protein kinase B, glycogen synthase kinase 3ß, and mammalian target of rapamycin are reported in both diseases. T2DM and AD also show evidence of inflammation, oxidative stress, mitochondrial dysfunction, advanced glycation end products, and amyloid deposition. The impact that changes in neurovascular structure and genetics have on the development of these conditions is also being examined. With the discovery of factors contributing to AD, innovative treatment approaches are being explored. Investigators are evaluating the efficacy of various T2DM medications for possible use in AD, including but not limited to glucagon-like peptide-1 receptor agonists and peroxisome proliferator-activated receptor-gamma agonists. Furthermore, there are 136 active trials involving 121 therapeutic agents targeting novel AD biomarkers. With these efforts, we are one step closer to alleviating the ravaging impact of AD on our communities.

Subcellular location and molecular mobility of human cytosolic sulfotransferase 1C1 in living human embryonic kidney 293 cells.

Cytosolic sulfotransferases were first isolated from the hepatic cytosol, and they have been localized in the cytoplasm of formaldehyde-fixed human cell samples. The current work was carried out to determine the subcellular localization and molecular mobility of cytosolic sulfotransferases in living human embryonic kidney (HEK) 293 cells. In this work, the subcellular location of human cytosolic sulfotransferase 1C1 (SULT1C1) was studied in cultured HEK293 cells using confocal laser-scanning microscopy. A green fluorescent protein (GFP)-tagged SULT1C1 protein was localized in the cytoplasm of living HEK293 cells. This is consistent with results from previous studies on several other cytosolic sulfotransferase isoforms. Fluorescence recovery after photobleaching microscopy was performed to assess the molecular mobility of the expressed GFP-SULT1C1 molecules. The results suggested that the expressed recombinant GFP-SULT1C1 molecules in living HEK293 cells may include both mobile and immobile populations. To obtain additional insights into the subcellular location of SULT1C1, two machine learning algorithms, Sequential Minimal Optimization and Multilayer Perceptron, were used to compute the probability distribution for the localization of SULT1C1 in nine selected cellular compartments. The resulting probability distribution suggested that the most likely subcellular location of SULT1C1 is the cytosol.

NRF2 Activation Promotes Aggressive Lung Cancer and Associates with Poor Clinical Outcomes.

PURPOSE: Stabilization of the transcription factor NRF2 through genomic alterations in KEAP1 and NFE2L2 occurs in a quarter of patients with lung adenocarcinoma and a third of patients with lung squamous cell carcinoma. In lung adenocarcinoma, KEAP1 loss often co-occurs with STK11 loss and KRAS-activating alterations. Despite its prevalence, the impact of NRF2 activation on tumor progression and patient outcomes is not fully defined. EXPERIMENTAL DESIGN: We model NRF2 activation, STK11 loss, and KRAS activation in vivo using novel genetically engineered mouse models. Furthermore, we derive a NRF2 activation signature from human non-small cell lung tumors that we use to dissect how these genomic events impact outcomes and immune contexture of participants in the OAK and IMpower131 immunotherapy trials. RESULTS: Our in vivo data reveal roles for NRF2 activation in (i) promoting rapid-onset, multifocal intrabronchiolar carcinomas, leading to lethal pulmonary dysfunction, and (ii) decreasing elevated redox stress in KRAS-mutant, STK11-null tumors. In patients with nonsquamous tumors, the NRF2 signature is negatively prognostic independently of STK11 loss. Patients with lung squamous cell carcinoma with low NRF2 signature survive longer when receiving anti-PD-L1 treatment. CONCLUSIONS: Our in vivo modeling establishes NRF2 activation as a critical oncogenic driver, cooperating with STK11 loss and KRAS activation to promote aggressive lung adenocarcinoma. In patients, oncogenic events alter the tumor immune contexture, possibly having an impact on treatment responses. Importantly, patients with NRF2-activated nonsquamous or squamous tumors have poor prognosis and show limited response to anti-PD-L1 treatment.

DOKI: Domain knowledge-driven inference method for reverse-engineering transcriptional regulatory relationships among genes in cancer.

Efficient reverse-engineering methods are important for identifying transcriptional regulatory relationships among genes in cancer. These methods are becoming increasingly useful in this era where huge volumes of data are generated through the use of high-throughput technologies such as next-generation sequencing technologies and microarrays. However, it is important to improve current methods because of complications involved in modelling complex biological systems. In this paper, we present a novel approach, Domain Knowledge-driven Inference (DOKI), for identification of transcriptional regulatory relationships among genes, given a biological context such as cancer. Combining data normalization, the use of a probability distribution function and Kullback-Leibler Divergence, DOKI incorporates a domain knowledge-driven criterion to make determinations of the existence of regulatory relationships between given transcription factors and given specific gene targets. Characteristics of DOKI enable it to adequately handle complexities inherent in data, and accurately unearth linear and higher-order dependent relationships among genes. DOKI performed equally well with one established high-performing method and better than three other high-performing methods on relatively small data sets. However, it remarkably outperformed these methods on larger data sets to demonstrate its utility. Furthermore, we demonstrate the relevance of such inference algorithms for identifying novel relationships among genes in breast cancer, as some of the consensus results representing novel relationships were confirmed in previously published experimental results. Thus, DOKI will facilitate current efforts to gain etiological insights and help uncover new targeted therapies for various diseases.

MSclassifier: median-supplement model-based classification tool for automated knowledge discovery.

High-throughput technologies have resulted in an exponential growth of publicly available and accessible datasets for biomedical research. Efficient computational models, algorithms and tools are required to exploit the datasets for knowledge discovery to aid medical decisions. Here, we introduce a new tool, MSclassifier, based on median-supplement approaches to machine learning to enable an automated and effective binary classification for optimal decision making. The MSclassifier package estimates medians of features (attributes) to deduce supplementary data, which is subsequently introduced into the training set for balancing and building superior models for classification. To test our approach, it is used to determine HER2 receptor expression status phenotypes in breast cancer and also predict protein subcellular localization (plasma membrane and nucleus). Using independent sample and cross-validation tests, the performance of MSclassifier is evaluated and compared with well established tools that could perform such tasks. In the HER2 receptor expression status phenotype identification tasks, MSclassifier achieved statistically significant higher classification rates than the best performing existing tool (90.30% versus 89.83%, p=8.62e-3). In the subcellular localization prediction tasks, MSclassifier and one other existing tool achieved equally high performances (93.42% versus 93.19%, p=0.06) although they both outperformed tools based on Naive Bayes classifiers. Overall, the application and evaluation of MSclassifier reveal its potential to be applied to varieties of binary classification problems. The MSclassifier package provides an R-portable and user-friendly application to a broad audience, enabling experienced end-users as well as non-programmers to perform an effective classification in biomedical and other fields of study.

Enhanced immortalization, HUWE1 mutations and other biological drivers of breast invasive carcinoma in Black/African American patients.

Black/African-American (B/AA) breast cancer patients tend to have more aggressive tumor biology compared to White/Caucasians. In this study, a variety of breast tumor molecular expression profiles of patients derived from the two racial groupings were investigated. Breast invasive carcinoma sample data (RNASeq version 2, Reverse Phase Protein Array, mutation, and miRSeq data) from the Cancer Genome Atlas were examined. The results affirm that B/AA patients are more likely than Caucasian patients to harbor the aggressive basal-like or the poor prognosis-associated HER2-enriched molecular subtypes of breast cancer. There is also a higher incidence of the triple-negative breast cancer (TNBC) among B/AA patients than the general population, a fact reflected in the mutation patterns of genes such as PIK3CA and TP53. Furthermore, an immortalization signature gene set, is enriched in samples from B/AA patients. Among stage III patients, TERT, DRAP1, and PQBP1, all members of the immortalization gene signature set, are among master-regulators with increased activity in B/AA patients. Master-regulators driving differences in expression profiles between the two groups include immortalization markers, senescence markers, and immune response and redox gene products. Differences in expression, between B/AA and Caucasian patients, of RB1, hsa-let-7a, E2F1, c-MYC, TERT, and other biomolecules appear to cooperate to enhance entry into the S-phase of the cell cycle in B/AA patients. Higher expression of miR-221, an oncomiR that facilitates entry into the cell cycle S-phase, is regulated by c-MYC, which is expressed more in breast cancer samples from B/AA patients. Furthermore, the cell migration- and invasion-promoting miRNA, miR-135b, has increased relative expression in B/AA patients. Knock down of the immortalization marker TERT inhibited triple-negative breast cancer cell lines (MDA-MB-231 and MDA-MB-468) cell viability and decreased expression of TERT, MYC and WNT11. For those patients with available survival data, prognosis of stage II patients 50 years of age or younger at diagnosis, was distinctly poorer in B/AA patients. Also associated with this subset of B/AA patients are missense mutations in HUWE1 and PTEN expression loss. Relative to Caucasian non-responders to endocrine therapy, B/AA non-responders show suppressed expression of a signature gene set on which biological processes including signaling by interleukins, circadian clock, regulation of lipid metabolism by PPARα, FOXO-mediated transcription, and regulation of TP53 degradation are over-represented. Thus, we identify molecular expression patterns suggesting diminished response to oxidative stress, changes in regulation of tumor suppressors/facilitators, and enhanced immortalization in B/AA patients are likely important in defining the more aggressive molecular tumor phenotype reported in B/AA patients.

Enhanced immortalization, HUWE1 mutations and other biological drivers of breast invasive carcinoma in Black/African American patients.

Black/African-American (B/AA) breast cancer patients tend to have more aggressive tumor biology compared to White/Caucasians. In this study, a variety of breast tumor molecular expression profiles of patients derived from the two racial groupings were investigated. Breast invasive carcinoma sample data (RNASeq version 2, Reverse Phase Protein Array, mutation, and miRSeq data) from the Cancer Genome Atlas were examined. The results affirm that B/AA patients are more likely than Caucasian patients to harbor the aggressive basal-like or the poor prognosis-associated HER2-enriched molecular subtypes of breast cancer. There is also a higher incidence of the triple-negative breast cancer (TNBC) among B/AA patients than the general population, a fact reflected in the mutation patterns of genes such as PIK3CA and TP53. Furthermore, an immortalization signature gene set, is enriched in samples from B/AA patients. Among stage III patients, TERT, DRAP1, and PQBP1, all members of the immortalization gene signature set, are among master-regulators with increased activity in B/AA patients. Master-regulators driving differences in expression profiles between the two groups include immortalization markers, senescence markers, and immune response and redox gene products. Differences in expression, between B/AA and Caucasian patients, of RB1, hsa-let-7a, E2F1, c-MYC, TERT, and other biomolecules appear to cooperate to enhance entry into the S-phase of the cell cycle in B/AA patients. Higher expression of miR-221, an oncomiR that facilitates entry into the cell cycle S-phase, is regulated by c-MYC, which is expressed more in breast cancer samples from B/AA patients. Furthermore, the cell migration- and invasion-promoting miRNA, miR-135b, has increased relative expression in B/AA patients. Knock down of the immortalization marker TERT inhibited triple-negative breast cancer cell lines (MDA-MB-231 and MDA-MB-468) cell viability and decreased expression of TERT, MYC and WNT11. For those patients with available survival data, prognosis of stage II patients 50 years of age or younger at diagnosis, was distinctly poorer in B/AA patients. Also associated with this subset of B/AA patients are missense mutations in HUWE1 and PTEN expression loss. Relative to Caucasian non-responders to endocrine therapy, B/AA non-responders show suppressed expression of a signature gene set on which biological processes including signaling by interleukins, circadian clock, regulation of lipid metabolism by PPARα, FOXO-mediated transcription, and regulation of TP53 degradation are over-represented. Thus, we identify molecular expression patterns suggesting diminished response to oxidative stress, changes in regulation of tumor suppressors/facilitators, and enhanced immortalization in B/AA patients are likely important in defining the more aggressive molecular tumor phenotype reported in B/AA patients.

Network inference algorithms elucidate Nrf2 regulation of mouse lung oxidative stress.

A variety of cardiovascular, neurological, and neoplastic conditions have been associated with oxidative stress, i.e., conditions under which levels of reactive oxygen species (ROS) are elevated over significant periods. Nuclear factor erythroid 2-related factor (Nrf2) regulates the transcription of several gene products involved in the protective response to oxidative stress. The transcriptional regulatory and signaling relationships linking gene products involved in the response to oxidative stress are, currently, only partially resolved. Microarray data constitute RNA abundance measures representing gene expression patterns. In some cases, these patterns can identify the molecular interactions of gene products. They can be, in effect, proxies for protein-protein and protein-DNA interactions. Traditional techniques used for clustering coregulated genes on high-throughput gene arrays are rarely capable of distinguishing between direct transcriptional regulatory interactions and indirect ones. In this study, newly developed information-theoretic algorithms that employ the concept of mutual information were used: the Algorithm for the Reconstruction of Accurate Cellular Networks (ARACNE), and Context Likelihood of Relatedness (CLR). These algorithms captured dependencies in the gene expression profiles of the mouse lung, allowing the regulatory effect of Nrf2 in response to oxidative stress to be determined more precisely. In addition, a characterization of promoter sequences of Nrf2 regulatory targets was conducted using a Support Vector Machine classification algorithm to corroborate ARACNE and CLR predictions. Inferred networks were analyzed, compared, and integrated using the Collective Analysis of Biological Interaction Networks (CABIN) plug-in of Cytoscape. Using the two network inference algorithms and one machine learning algorithm, a number of both previously known and novel targets of Nrf2 transcriptional activation were identified. Genes predicted as novel Nrf2 targets include Atf1, Srxn1, Prnp, Sod2, Als2, Nfkbib, and Ppp1r15b. Furthermore, microarray and quantitative RT-PCR experiments following cigarette-smoke-induced oxidative stress in Nrf2(+/+) and Nrf2(-/-) mouse lung affirmed many of the predictions made. Several new potential feed-forward regulatory loops involving Nrf2, Nqo1, Srxn1, Prdx1, Als2, Atf1, Sod1, and Park7 were predicted. This work shows the promise of network inference algorithms operating on high-throughput gene expression data in identifying transcriptional regulatory and other signaling relationships implicated in mammalian disease.

Restricted-derestricted dynamic Bayesian Network inference of transcriptional regulatory relationships among genes in cancer.

Understanding transcriptional regulatory relationships among genes is important for gaining etiological insights into diseases such as cancer. To this end, high-throughput biological data have been generated through advancements in a variety of technologies. These rely on computational approaches to discover underlying structures in such data. Among these computational approaches, Bayesian networks (BNs) stand out because their probabilistic nature enables them to manage randomness in the dynamics of gene regulation and experimental data. Feedback loops inherent in networks of regulatory relationships are more tractable when enhancements to BNs are applied to them. Here, we propose Restricted-Derestricted dynamic BNs with a novel search technique, Restricted-Derestricted Greedy Method, for such tasks. This approach relies on the Restricted-Derestricted Greedy search technique to infer transcriptional regulatory networks in two phases: restricted inference and derestricted inference. An application of this approach to real data sets reveals it performs favourably well compared to other existing well performing dynamic BN approaches in terms of recovering true relationships among genes. In addition, it provides a balance between searching for optimal networks and keeping biologically relevant regulatory interactions among variables.

Manual curation is not sufficient for annotation of genomic databases.

MOTIVATION: Knowledge base construction has been an area of intense activity and great importance in the growth of computational biology. However, there is little or no history of work on the subject of evaluation of knowledge bases, either with respect to their contents or with respect to the processes by which they are constructed. This article proposes the application of a metric from software engineering known as the found/fixed graph to the problem of evaluating the processes by which genomic knowledge bases are built, as well as the completeness of their contents. RESULTS: Well-understood patterns of change in the found/fixed graph are found to occur in two large publicly available knowledge bases. These patterns suggest that the current manual curation processes will take far too long to complete the annotations of even just the most important model organisms, and that at their current rate of production, they will never be sufficient for completing the annotation of all currently available proteomes.

Ethanol sensitivity: a central role for CREB transcription regulation in the cerebellum.

BACKGROUND: Lowered sensitivity to the effects of ethanol increases the risk of developing alcoholism. Inbred mouse strains have been useful for the study of the genetic basis of various drug addiction-related phenotypes. Inbred Long-Sleep (ILS) and Inbred Short-Sleep (ISS) mice differentially express a number of genes thought to be implicated in sensitivity to the effects of ethanol. Concomitantly, there is evidence for a mediating role of cAMP/PKA/CREB signalling in aspects of alcoholism modelled in animals. In this report, the extent to which CREB signalling impacts the differential expression of genes in ILS and ISS mouse cerebella is examined. RESULTS: A training dataset for Machine Learning (ML) and Exploratory Data Analyses (EDA) was generated from promoter region sequences of a set of genes known to be targets of CREB transcription regulation and a set of genes whose transcription regulations are potentially CREB-independent. For each promoter sequence, a vector of size 132, with elements characterizing nucleotide composition features was generated. Genes whose expressions have been previously determined to be increased in ILS or ISS cerebella were identified, and their CREB regulation status predicted using the ML scheme C4.5. The C4.5 learning scheme was used because, of four ML schemes evaluated, it had the lowest predicted error rate. On an independent evaluation set of 21 genes of known CREB regulation status, C4.5 correctly classified 81% of instances with F-measures of 0.87 and 0.67 respectively for the CREB-regulated and CREB-independent classes. Additionally, six out of eight genes previously determined by two independent microarray platforms to be up-regulated in the ILS or ISS cerebellum were predicted by C4.5 to be transcriptionally regulated by CREB. Furthermore, 64% and 52% of a cross-section of other up-regulated cerebellar genes in ILS and ISS mice, respectively, were deemed to be CREB-regulated. CONCLUSION: These observations collectively suggest that ethanol sensitivity, as it relates to the cerebellum, may be associated with CREB transcription activity.

Predicting the subcellular localization of human proteins using machine learning and exploratory data analysis.

Identifying the subcellular localization of proteins is particularly helpful in the functional annotation of gene products. In this study, we use Machine Learning and Exploratory Data Analysis (EDA) techniques to examine and characterize amino acid sequences of human proteins localized in nine cellular compartments. A dataset of 3,749 protein sequences representing human proteins was extracted from the SWISS-PROT database. Feature vectors were created to capture specific amino acid sequence characteristics. Relative to a Support Vector Machine, a Multi-layer Perceptron, and a Naive Bayes classifier, the C4.5 Decision Tree algorithm was the most consistent performer across all nine compartments in reliably predicting the subcellular localization of proteins based on their amino acid sequences (average Precision=0.88; average Sensitivity=0.86). Furthermore, EDA graphics characterized essential features of proteins in each compartment. As examples, proteins localized on the plasma membrane had higher proportions of hydrophobic amino acids; cytoplasmic proteins had higher proportions of neutral amino acids; and mitochondrial proteins had higher proportions of neutral amino acids and lower proportions of polar amino acids. These data showed that the C4.5 classifier and EDA tools can be effective for characterizing and predicting the subcellular localization of human proteins based on their amino acid sequences.

Brain in situ hybridization maps as a source for reverse-engineering transcriptional regulatory networks: Alzheimer's disease insights.

Microarray data have been a valuable resource for identifying transcriptional regulatory relationships among genes. As an example, brain region-specific transcriptional regulatory events have the potential of providing etiological insights into Alzheimer Disease (AD). However, there is often a paucity of suitable brain-region specific expression data obtained via microarrays or other high throughput means. The Allen Brain Atlas in situ hybridization (ISH) data sets (Jones et al., 2009) represent a potentially valuable alternative source of high-throughput brain region-specific gene expression data for such purposes. In this study, Allen Brain Atlas mouse ISH data in the hippocampal fields were extracted, focusing on 508 genes relevant to neurodegeneration. Transcriptional regulatory networks were learned using three high-performing network inference algorithms. Only 17% of regulatory edges from a network reverse-engineered based on brain region-specific ISH data were also found in a network constructed upon gene expression correlations in mouse whole brain microarrays, thus showing the specificity of gene expression within brain sub-regions. Furthermore, the ISH data-based networks were used to identify instructive transcriptional regulatory relationships. Ncor2, Sp3 and Usf2 form a unique three-party regulatory motif, potentially affecting memory formation pathways. Nfe2l1, Egr1 and Usf2 emerge among regulators of genes involved in AD (e.g. Dhcr24, Aplp2, Tia1, Pdrx1, Vdac1, and Syn2). Further, Nfe2l1, Egr1 and Usf2 are sensitive to dietary factors and could be among links between dietary influences and genes in the AD etiology. Thus, this approach of harnessing brain region-specific ISH data represents a rare opportunity for gleaning unique etiological insights for diseases such as AD.