Correcting Smad1/5/8, mTOR, and VEGFR2 treats pathology in hereditary hemorrhagic telangiectasia models.

Hereditary hemorrhagic telangiectasia (HHT), a genetic bleeding disorder leading to systemic arteriovenous malformations (AVMs), is caused by loss-of-function mutations in the ALK1/ENG/Smad1/5/8 pathway. Evidence suggests that HHT pathogenesis strongly relies on overactivated PI3K/Akt/mTOR and VEGFR2 pathways in endothelial cells (ECs). In the BMP9/10-immunoblocked (BMP9/10ib) neonatal mouse model of HHT, we report here that the mTOR inhibitor, sirolimus, and the receptor tyrosine kinase inhibitor, nintedanib, could synergistically fully block, but also reversed, retinal AVMs to avert retinal bleeding and anemia. Sirolimus plus nintedanib prevented vascular pathology in the oral mucosa, lungs, and liver of the BMP9/10ib mice, as well as significantly reduced gastrointestinal bleeding and anemia in inducible ALK1-deficient adult mice. Mechanistically, in vivo in BMP9/10ib mouse ECs, sirolimus and nintedanib blocked the overactivation of mTOR and VEGFR2, respectively. Furthermore, we found that sirolimus activated ALK2-mediated Smad1/5/8 signaling in primary ECs - including in HHT patient blood outgrowth ECs - and partially rescued Smad1/5/8 activity in vivo in BMP9/10ib mouse ECs. These data demonstrate that the combined correction of endothelial Smad1/5/8, mTOR, and VEGFR2 pathways opposes HHT pathogenesis. Repurposing of sirolimus plus nintedanib might provide therapeutic benefit in patients with HHT.

Tacrolimus rescues the signaling and gene expression signature of endothelial ALK1 loss-of-function and improves HHT vascular pathology.

Hereditary hemorrhagic telangiectasia (HHT) is a highly debilitating and life-threatening genetic vascular disorder arising from endothelial cell (EC) proliferation and hypervascularization, for which no cure exists. Because HHT is caused by loss-of-function mutations in bone morphogenetic protein 9 (BMP9)-ALK1-Smad1/5/8 signaling, interventions aimed at activating this pathway are of therapeutic value. We interrogated the whole-transcriptome in human umbilical vein ECs (HUVECs) and found that ALK1 signaling inhibition was associated with a specific pro-angiogenic gene expression signature, which included a significant elevation of DLL4 expression. By screening the NIH clinical collections of FDA-approved drugs, we identified tacrolimus (FK-506) as the most potent activator of ALK1 signaling in BMP9-challenged C2C12 reporter cells. In HUVECs, tacrolimus activated Smad1/5/8 and opposed the pro-angiogenic gene expression signature associated with ALK1 loss-of-function, by notably reducing Dll4 expression. In these cells, tacrolimus also inhibited Akt and p38 stimulation by vascular endothelial growth factor, a major driver of angiogenesis. In the BMP9/10-immunodepleted postnatal retina-a mouse model of HHT vascular pathology-tacrolimus activated endothelial Smad1/5/8 and prevented the Dll4 overexpression and hypervascularization associated with this model. Finally, tacrolimus stimulated Smad1/5/8 signaling in C2C12 cells expressing BMP9-unresponsive ALK1 HHT mutants and in HHT patient blood outgrowth ECs. Tacrolimus repurposing has therefore therapeutic potential in HHT.

Specific calpain inhibition protects kidney against inflammaging.

Calpains are ubiquitous pro-inflammatory proteases, whose activity is controlled by calpastatin, their specific inhibitor. Transgenic mice over-expressing rabbit calpastatin (CalpTG) are protected against vascular remodelling and angiotensin II-dependent inflammation. We hypothesized that specific calpain inhibition would protect against aging-related lesions in arteries and kidneys. We analysed tissues from 2-months and 2-years-old CalpTG and wild-type mice and performed high throughput RNA-Sequencing of kidney tissue in aged mice. In addition, we analysed inflammatory response in the kidney of aged CalpTG and wild-type mice, and in both in vivo (monosodium urate peritonitis) and in vitro models of inflammation. At two years, CalpTG mice had preserved kidney tissue, less vascular remodelling and less markers of senescence than wild-type mice. Nevertheless, CalpTG mice lifespan was not extended, due to the development of lethal spleen tumors. Inflammatory pathways were less expressed in aged CalpTG mice, especially cytokines related to NF-κB and NLRP3 inflammasome activation. CalpTG mice had reduced macrophage infiltration with aging and CalpTG mice produced less IL-1α and IL-1ß in vivo in response to inflammasome activators. In vitro, macrophages from CalpTG mice produced less IL-1α in response to particulate activators of inflammasome. Calpains inhibition protects against inflammaging, limiting kidney and vascular lesions related to aging.

Glomerular common gamma chain confers B- and T-cell-independent protection against glomerulonephritis.

Crescentic glomerulonephritis is a life-threatening renal disease that has been extensively studied by the experimental anti-glomerular basement membrane glomerulonephritis (anti-GBM-GN) model. Although T cells have a significant role in this model, athymic/nude mice and rats still develop severe renal disease. Here we further explored the contribution of intrinsic renal cells in the development of T-cell-independent GN lesions. Anti-GBM-GN was induced in three strains of immune-deficient mice (Rag2-/-, Rag2-/-Il2rg-/-, and Rag2-/-Il2rb-/-) that are devoid of either T/B cells or T/B/NK cells. The Rag2-/-Il2rg-/- or Rag2-/-Il2rb-/- mice harbor an additional deletion of either the common gamma chain (γC) or the interleukin-2 receptor ß subunit (IL-2Rß), respectively, impairing IL-15 signaling in particular. As expected, all these strains developed severe anti-GBM-GN. Additionally, bone marrow replenishment experiments allowed us to deduce a protective role for the glomerular-expressed γC during anti-GBM-GN. Given that IL-15 has been found highly expressed in nephritic kidneys despite the absence of lymphocytes, we then studied this cytokine in vitro on primary cultured podocytes from immune-deficient mice (Rag2-/-Il2rg-/- and Rag2-/-Il2rb-/-) compared to controls. IL-15 induced downstream activation of JAK1/3 and SYK in primary cultured podocytes. IL-15-dependent JAK/SYK induction was impaired in the absence of γC or IL-2Rß. We found γC largely induced on podocytes during human glomerulonephritis. Thus, renal lesions are indeed modulated by intrinsic glomerular cells through the γC/IL-2Rß receptor response, to date classically described only in immune cells.

A mouse model of hereditary hemorrhagic telangiectasia generated by transmammary-delivered immunoblocking of BMP9 and BMP10.

Hereditary hemorrhagic telangiectasia (HHT) is a potentially life-threatening genetic vascular disorder caused by loss-of-function mutations in the genes encoding activin receptor-like kinase 1 (ALK1), endoglin, Smad4, and bone morphogenetic protein 9 (BMP9). Injections of mouse neonates with BMP9/10 blocking antibodies lead to HHT-like vascular defects in the postnatal retinal angiogenesis model. Mothers and their newborns share the same immunity through the transfer of maternal antibodies during lactation. Here, we investigated whether the transmammary delivery route could improve the ease and consistency of administering anti-BMP9/10 antibodies in the postnatal retinal angiogenesis model. We found that anti-BMP9/10 antibodies, when intraperitoneally injected into lactating dams, are efficiently transferred into the blood circulation of lactationally-exposed neonatal pups. Strikingly, pups receiving anti-BMP9/10 antibodies via lactation displayed consistent and robust vascular pathology in the retina, which included hypervascularization and defects in arteriovenous specification, as well as the presence of multiple and massive arteriovenous malformations. Furthermore, RNA-Seq analyses of neonatal retinas identified an increase in the key pro-angiogenic factor, angiopoietin-2, as the most significant change in gene expression triggered by the transmammary delivery of anti-BMP9/10 antibodies. Transmammary-delivered BMP9/10 immunoblocking in the mouse neonatal retina is therefore a practical, noninvasive, reliable, and robust model of HHT vascular pathology.

Exome Sequencing and Prediction of Long-Term Kidney Allograft Function.

Current strategies to improve graft outcome following kidney transplantation consider information at the human leukocyte antigen (HLA) loci. Cell surface antigens, in addition to HLA, may serve as the stimuli as well as the targets for the anti-allograft immune response and influence long-term graft outcomes. We therefore performed exome sequencing of DNA from kidney graft recipients and their living donors and estimated all possible cell surface antigens mismatches for a given donor/recipient pair by computing the number of amino acid mismatches in trans-membrane proteins. We designated this tally as the allogenomics mismatch score (AMS). We examined the association between the AMS and post-transplant estimated glomerular filtration rate (eGFR) using mixed models, considering transplants from three independent cohorts (a total of 53 donor-recipient pairs, 106 exomes, and 239 eGFR measurements). We found that the AMS has a significant effect on eGFR (mixed model, effect size across the entire range of the score: -19.4 [-37.7, -1.1], P = 0.0042, χ2 = 8.1919, d.f. = 1) that is independent of the HLA-A, B, DR matching, donor age, and time post-transplantation. The AMS effect is consistent across the three independent cohorts studied and similar to the strong effect size of donor age. Taken together, these results show that the AMS, a novel tool to quantify amino acid mismatches in trans-membrane proteins in individual donor/recipient pair, is a strong, robust predictor of long-term graft function in kidney transplant recipients.

Whole-Exome Sequencing of Metastatic Cancer and Biomarkers of Treatment Response.

IMPORTANCE: Understanding molecular mechanisms of response and resistance to anticancer therapies requires prospective patient follow-up and clinical and functional validation of both common and low-frequency mutations. We describe a whole-exome sequencing (WES) precision medicine trial focused on patients with advanced cancer. OBJECTIVE: To understand how WES data affect therapeutic decision making in patients with advanced cancer and to identify novel biomarkers of response. DESIGN, SETTING, AND PATIENTS: Patients with metastatic and treatment-resistant cancer were prospectively enrolled at a single academic center for paired metastatic tumor and normal tissue WES during a 19-month period (February 2013 through September 2014). A comprehensive computational pipeline was used to detect point mutations, indels, and copy number alterations. Mutations were categorized as category 1, 2, or 3 on the basis of actionability; clinical reports were generated and discussed in precision tumor board. Patients were observed for 7 to 25 months for correlation of molecular information with clinical response. MAIN OUTCOMES AND MEASURES: Feasibility, use of WES for decision making, and identification of novel biomarkers. RESULTS: A total of 154 tumor-normal pairs from 97 patients with a range of metastatic cancers were sequenced, with a mean coverage of 95X and 16 somatic alterations detected per patient. In total, 16 mutations were category 1 (targeted therapy available), 98 were category 2 (biologically relevant), and 1474 were category 3 (unknown significance). Overall, WES provided informative results in 91 cases (94%), including alterations for which there is an approved drug, there are therapies in clinical or preclinical development, or they are considered drivers and potentially actionable (category 1-2); however, treatment was guided in only 5 patients (5%) on the basis of these recommendations because of access to clinical trials and/or off-label use of drugs. Among unexpected findings, a patient with prostate cancer with exceptional response to treatment was identified who harbored a somatic hemizygous deletion of the DNA repair gene FANCA and putative partial loss of function of the second allele through germline missense variant. Follow-up experiments established that loss of FANCA function was associated with platinum hypersensitivity both in vitro and in patient-derived xenografts, thus providing biologic rationale and functional evidence for his extreme clinical response. CONCLUSIONS AND RELEVANCE: The majority of advanced, treatment-resistant tumors across tumor types harbor biologically informative alterations. The establishment of a clinical trial for WES of metastatic tumors with prospective follow-up of patients can help identify candidate predictive biomarkers of response.

In Silico Analysis of NF2 Gene Missense Mutations in Neurofibromatosis Type 2: From Genotype to Phenotype.

HYPOTHESIS: Computer-based (in silico) protein modeling to examine genotype-phenotype relationships for a given mutation has been applied to many genes but never to NF2. BACKGROUND: Missense mutations in the merlin protein occur in approximately 9% of patients with neurofibromatosis type 2 (NF2). Within this subset of patients, no genotype-phenotype correlations have been established. The aim of this study was to determine if genotype correlates with phenotype in the cohort of NF2 patients with missense mutations as a first step to defining a method to predict clinical phenotype from genotype for these patients. METHODS: We analyzed 45 patients with NF2 as a result of missense mutations drawn from the United Kingdom NF2 registry. Our analysis included 17 different NF2 mutations from NF2 patients and six single-nucleotide polymorphisms (SNP)--presumed benign because they are observed in the dbSNP National Center for Biotechnology Information database and 1000 Genomes. We analyzed the mutations using three mutation tolerance prediction approaches: Align GVGD, SIFT, and PolyPhen-2. The mutation sites were also modeled on the three-dimensional crystal structure of merlin to investigate the spatial relationship of NF2-causing mutations. RESULTS: Two mutation tolerance predictors (SIFT and PolyPhen-2) were able to distinguish NF2-causing mutations from non-NF2-causing SNPs (p < 0.05). Mapping mutations on the molecular structure of merlin suggest that mutations resulting in greater structural conflicts within the protein are more likely to correlate with severe phenotypes. CONCLUSION: This work is a step toward a better understanding of genotype-phenotype relationships in NF2 caused by missense mutations using a computer-based methodology.

Language workbench user interfaces for data analysis.

Biological data analysis is frequently performed with command line software. While this practice provides considerable flexibility for computationally savy individuals, such as investigators trained in bioinformatics, this also creates a barrier to the widespread use of data analysis software by investigators trained as biologists and/or clinicians. Workflow systems such as Galaxy and Taverna have been developed to try and provide generic user interfaces that can wrap command line analysis software. These solutions are useful for problems that can be solved with workflows, and that do not require specialized user interfaces. However, some types of analyses can benefit from custom user interfaces. For instance, developing biomarker models from high-throughput data is a type of analysis that can be expressed more succinctly with specialized user interfaces. Here, we show how Language Workbench (LW) technology can be used to model the biomarker development and validation process. We developed a language that models the concepts of Dataset, Endpoint, Feature Selection Method and Classifier. These high-level language concepts map directly to abstractions that analysts who develop biomarker models are familiar with. We found that user interfaces developed in the Meta-Programming System (MPS) LW provide convenient means to configure a biomarker development project, to train models and view the validation statistics. We discuss several advantages of developing user interfaces for data analysis with a LW, including increased interface consistency, portability and extension by language composition. The language developed during this experiment is distributed as an MPS plugin (available at http://campagnelab.org/software/bdval-for-mps/).

Effect of the CALHM1 G330D and R154H human variants on the control of cytosolic Ca2+ and Aß levels.

CALHM1 is a plasma membrane voltage-gated Ca2+-permeable ion channel that controls amyloid-ß (Aß) metabolism and is potentially involved in the onset of Alzheimer's disease (AD). Recently, Rubio-Moscardo et al. (PLoS One (2013) 8: e74203) reported the identification of two CALHM1 variants, G330D and R154H, in early-onset AD (EOAD) patients. The authors provided evidence that these two human variants were rare and resulted in a complete loss of CALHM1 function. Recent publicly available large-scale exome sequencing data confirmed that R154H is a rare CALHM1 variant (minor allele frequency (MAF)  = 0.015%), but that G330D is not (MAF  = 3.5% in an African American cohort). Here, we show that both CALHM1 variants exhibited gating and permeation properties indistinguishable from wild-type CALHM1 when expressed in Xenopus oocytes. While there was also no effect of the G330D mutation on Ca2+ uptake by CALHM1 in transfected mammalian cells, the R154H mutation was associated with defects in the control by CALHM1 of both Ca2+ uptake and Aß levels in this cell system. Together, our data show that the frequent CALHM1 G330D variant has no obvious functional consequences and is therefore unlikely to contribute to EOAD. Our data also demonstrate that the rare R154H variant interferes with CALHM1 control of cytosolic Ca2+ and Aß accumulation. While these results strengthen the notion that CALHM1 influences Aß metabolism, further investigation will be required to determine whether CALHM1 R154H, or other natural variants in CALHM1, is/are associated with EOAD.

Composable languages for bioinformatics: the NYoSh experiment.

Language WorkBenches (LWBs) are software engineering tools that help domain experts develop solutions to various classes of problems. Some of these tools focus on non-technical users and provide languages to help organize knowledge while other workbenches provide means to create new programming languages. A key advantage of language workbenches is that they support the seamless composition of independently developed languages. This capability is useful when developing programs that can benefit from different levels of abstraction. We reasoned that language workbenches could be useful to develop bioinformatics software solutions. In order to evaluate the potential of language workbenches in bioinformatics, we tested a prominent workbench by developing an alternative to shell scripting. To illustrate what LWBs and Language Composition can bring to bioinformatics, we report on our design and development of NYoSh (Not Your ordinary Shell). NYoSh was implemented as a collection of languages that can be composed to write programs as expressive and concise as shell scripts. This manuscript offers a concrete illustration of the advantages and current minor drawbacks of using the MPS LWB. For instance, we found that we could implement an environment-aware editor for NYoSh that can assist the programmers when developing scripts for specific execution environments. This editor further provides semantic error detection and can be compiled interactively with an automatic build and deployment system. In contrast to shell scripts, NYoSh scripts can be written in a modern development environment, supporting context dependent intentions and can be extended seamlessly by end-users with new abstractions and language constructs. We further illustrate language extension and composition with LWBs by presenting a tight integration of NYoSh scripts with the GobyWeb system. The NYoSh Workbench prototype, which implements a fully featured integrated development environment for NYoSh is distributed at http://nyosh.campagnelab.org.

Variability in DNA methylation defines novel epigenetic subgroups of DLBCL associated with different clinical outcomes.

Diffuse large B-cell lymphoma (DLBCL) is the most common aggressive form of non-Hodgkin lymphoma with variable biology and clinical behavior. The current classification does not fully explain the biological and clinical heterogeneity of DLBCLs. In this study, we carried out genomewide DNA methylation profiling of 140 DLBCL samples and 10 normal germinal center B cells using the HpaII tiny fragment enrichment by ligation-mediated polymerase chain reaction assay and hybridization to a custom Roche NimbleGen promoter array. We defined methylation disruption as a main epigenetic event in DLBCLs and designed a method for measuring the methylation variability of individual cases. We then used a novel approach for unsupervised hierarchical clustering based on the extent of DNA methylation variability. This approach identified 6 clusters (A-F). The extent of methylation variability was associated with survival outcomes, with significant differences in overall and progression-free survival. The novel clusters are characterized by disruption of specific biological pathways such as cytokine-mediated signaling, ephrin signaling, and pathways associated with apoptosis and cell-cycle regulation. In a subset of patients, we profiled gene expression and genomic variation to investigate their interplay with methylation changes. This study is the first to identify novel epigenetic clusters of DLBCLs and their aberrantly methylated genes, molecular associations, and survival.

Compression of structured high-throughput sequencing data.

Large biological datasets are being produced at a rapid pace and create substantial storage challenges, particularly in the domain of high-throughput sequencing (HTS). Most approaches currently used to store HTS data are either unable to quickly adapt to the requirements of new sequencing or analysis methods (because they do not support schema evolution), or fail to provide state of the art compression of the datasets. We have devised new approaches to store HTS data that support seamless data schema evolution and compress datasets substantially better than existing approaches. Building on these new approaches, we discuss and demonstrate how a multi-tier data organization can dramatically reduce the storage, computational and network burden of collecting, analyzing, and archiving large sequencing datasets. For instance, we show that spliced RNA-Seq alignments can be stored in less than 4% the size of a BAM file with perfect data fidelity. Compared to the previous compression state of the art, these methods reduce dataset size more than 40% when storing exome, gene expression or DNA methylation datasets. The approaches have been integrated in a comprehensive suite of software tools (http://goby.campagnelab.org) that support common analyses for a range of high-throughput sequencing assays.

GobyWeb: simplified management and analysis of gene expression and DNA methylation sequencing data.

We present GobyWeb, a web-based system that facilitates the management and analysis of high-throughput sequencing (HTS) projects. The software provides integrated support for a broad set of HTS analyses and offers a simple plugin extension mechanism. Analyses currently supported include quantification of gene expression for messenger and small RNA sequencing, estimation of DNA methylation (i.e., reduced bisulfite sequencing and whole genome methyl-seq), or the detection of pathogens in sequenced data. In contrast to previous analysis pipelines developed for analysis of HTS data, GobyWeb requires significantly less storage space, runs analyses efficiently on a parallel grid, scales gracefully to process tens or hundreds of multi-gigabyte samples, yet can be used effectively by researchers who are comfortable using a web browser. We conducted performance evaluations of the software and found it to either outperform or have similar performance to analysis programs developed for specialized analyses of HTS data. We found that most biologists who took a one-hour GobyWeb training session were readily able to analyze RNA-Seq data with state of the art analysis tools. GobyWeb can be obtained at http://gobyweb.campagnelab.org and is freely available for non-commercial use. GobyWeb plugins are distributed in source code and licensed under the open source LGPL3 license to facilitate code inspection, reuse and independent extensions http://github.com/CampagneLaboratory/gobyweb2-plugins.

CALHM1 P86L polymorphism modulates CSF Aß levels in cognitively healthy individuals at risk for Alzheimer's disease.

The calcium homeostasis modulator 1 (CALHM1) gene codes for a novel cerebral calcium channel controlling intracellular calcium homeostasis and amyloid-ß (Aß) peptide metabolism, a key event in the etiology of Alzheimer's disease (AD). The P86L polymorphism in CALHM1 (rs2986017) initially was proposed to impair CALHM1 functionally and to lead to an increase in Aß accumulation in vitro in cell lines. Recently, it was reported that CALHM1 P86L also may influence Aß metabolism in vivo by increasing Aß levels in human cerebrospinal fluid (CSF). Although the role of CALHM1 in AD risk remains uncertain, concordant data have now emerged showing that CALHM1 P86L is associated with an earlier age at onset of AD. Here, we have analyzed the association of CALHM1 P86L with CSF Aß in samples from 203 AD cases and 46 young cognitively healthy individuals with a positive family history of AD. We failed to detect an association between the CALHM1 polymorphism and CSF Aß levels in AD patients. Our data, however, revealed a significant association of CALHM1 P86L with elevated CSF Aß42 and Aß40 in the normal cohort at risk for AD. This work shows that CALHM1 modulates CSF Aß levels in presymptomatic individuals, strengthening the notion that CALHM1 is involved in AD pathogenesis. These data further demonstrate the utility of endophenotype-based approaches focusing on CSF biomarkers for the identification or validation of risk factors for AD.

Growth arrest-specific 1 binds to and controls the maturation and processing of the amyloid-beta precursor protein.

Alzheimer's disease (AD), the most common neurodegenerative disorder, is characterized by cerebral deposition of amyloid-ß (Aß), a series of peptides derived from the processing of the amyloid-ß precursor protein (APP). To identify new candidate genes for AD, we recently performed a transcriptome analysis to screen for genes preferentially expressed in the hippocampus and located in AD linkage regions. This strategy identified CALHM1 (calcium homeostasis modulator 1), a gene modulating AD age at onset and Aß metabolism. Here, we focused our attention on another candidate identified using this screen, growth arrest-specific 1 (Gas1), a gene involved in the central nervous system development. We found that Gas1 formed a complex with APP and controlled APP maturation and processing. Gas1 expression inhibited APP full glycosylation and routing to the cell surface by leading to a trafficking blockade of APP between the endoplasmic reticulum and the Golgi. Gas1 expression also resulted in a robust inhibition of APP transport into multivesicular bodies, further demonstrating that Gas1 negatively regulated APP intracellular trafficking. Consequently, Gas1 overexpression led to a reduction in Aß production, and conversely, Gas1 silencing in cells expressing endogenously Gas1 increased Aß levels. These results suggest that Gas1 is a novel APP-interacting protein involved in the control of APP maturation and processing.

BDVal: reproducible large-scale predictive model development and validation in high-throughput datasets.

UNLABELLED: High-throughput data can be used in conjunction with clinical information to develop predictive models. Automating the process of developing, evaluating and testing such predictive models on different datasets would minimize operator errors and facilitate the comparison of different modeling approaches on the same dataset. Complete automation would also yield unambiguous documentation of the process followed to develop each model. We present the BDVal suite of programs that fully automate the construction of predictive classification models from high-throughput data and generate detailed reports about the model construction process. We have used BDVal to construct models from microarray and proteomics data, as well as from DNA-methylation datasets. The programs are designed for scalability and support the construction of thousands of alternative models from a given dataset and prediction task. AVAILABILITY AND IMPLEMENTATION: The BDVal programs are implemented in Java, provided under the GNU General Public License and freely available at http://bdval.campagnelab.org.

The MicroArray Quality Control (MAQC)-II study of common practices for the development and validation of microarray-based predictive models.

Gene expression data from microarrays are being applied to predict preclinical and clinical endpoints, but the reliability of these predictions has not been established. In the MAQC-II project, 36 independent teams analyzed six microarray data sets to generate predictive models for classifying a sample with respect to one of 13 endpoints indicative of lung or liver toxicity in rodents, or of breast cancer, multiple myeloma or neuroblastoma in humans. In total, >30,000 models were built using many combinations of analytical methods. The teams generated predictive models without knowing the biological meaning of some of the endpoints and, to mimic clinical reality, tested the models on data that had not been used for training. We found that model performance depended largely on the endpoint and team proficiency and that different approaches generated models of similar performance. The conclusions and recommendations from MAQC-II should be useful for regulatory agencies, study committees and independent investigators that evaluate methods for global gene expression analysis.

Introduction to the development and validation of predictive biomarker models from high-throughput data sets.

High-throughput technologies can routinely assay biological or clinical samples and produce wide data sets where each sample is associated with tens of thousands of measurements. Such data sets can be mined to discover biomarkers and develop statistical models capable of predicting an endpoint of interest from data measured in the samples. The field of biomarker model development combines methods from statistics and machine learning to develop and evaluate predictive biomarker models. In this chapter, we discuss the computational steps involved in the development of biomarker models designed to predict information about individual samples and review approaches often used to implement each step. A practical example of biomarker model development in a large gene expression data set is presented. This example leverages BDVal, a suite of biomarker model development programs developed as an open-source project (see http://bdval.org /).

GPCR-OKB: the G Protein Coupled Receptor Oligomer Knowledge Base.

SUMMARY: Rapid expansion of available data about G Protein Coupled Receptor (GPCR) dimers/oligomers over the past few years requires an effective system to organize this information electronically. Based on an ontology derived from a community dialog involving colleagues using experimental and computational methodologies, we developed the GPCR-Oligomerization Knowledge Base (GPCR-OKB). GPCR-OKB is a system that supports browsing and searching for GPCR oligomer data. Such data were manually derived from the literature. While focused on GPCR oligomers, GPCR-OKB is seamlessly connected to GPCRDB, facilitating the correlation of information about GPCR protomers and oligomers. AVAILABILITY AND IMPLEMENTATION: The GPCR-OKB web application is freely available at http://www.gpcr-okb.org