Search

Saltpans are increasingly being abandoned around the world, leading to the loss of ecosystem services provided by these unique semi-natural wetlands. The desertion of a saltpan at the Swartkops Estuary, South Africa has left behind a large area of desiccated hypersaline sediment and a sharp decrease in waterbird abundance and diversity. Here, we explore the potential for restoring this saltpan’s wetland function using stormwater inflow to support multiple ecosystem services and improve estuary health. Stormwater will be able to flow into the saltpan as a passive restoration approach that can maintain the site as a wetland habitat. This will contribute to improving the health of a nationally important, yet highly degraded, estuary by retaining 758 ML of stormwater to achieve full capacity, containing an estimated 2908 kg of dissolved inorganic nitrogen and 68 kg of dissolved inorganic nitrogen. Additional ecosystem services such as biodiversity maintenance, carbon storage and societal values can be expected. However, this would create a novel hypersaline stormwater wetland and a strategic adaptive management approach will be required. Management must be guided by monitoring, which should comprise collection of basic environmental data to establish a baseline condition, and against which, long term changes and responses resulting from stochastic events can be assessed and mitigated through the use of achievable ecological restoration targets.

This research has two goals: to measure the extent of assortative (non-random) mating by college major in the United States, and to assess the extent to which assortative mating by college major increases earnings inequality among college-educated couples. Assortative mating of college graduates with other college graduates has been extensively studied, but research on assortative mating by field of study is rare. The analysis uses a large sample (659,732 couples) from five years of the American Community Survey public use files to group college degrees into nine categories, compute the frequency of all marital pairings, and compare these frequencies to a random assignment of pairings. The results show that assortative mating by college major is common for all majors and both genders, and that these results are robust to division of the sample by age group. Because high-earning majors tend to be married to spouses from the same high-earning major group, and likewise for low-earning majors, assortative mating increases earnings inequality among two-earner college-educated couples. The extent of this increased earnings inequality is calculated with both dollar measures and standard aggregate measures of inequality. Thus college-educated Americans tend to marry persons with similar college majors and this tendency measurably increases earnings inequality among college-educated couples.

High-fidelity numerical tools based on high-order Discontinuous-Galerkin (DG) methods and Lagrangian Coherent Structure (LCS) theory are developed and validated for the study of separated, vortex-dominated flows over complex geometry. The numerical framework couples prediction of separated turbulent flows using DG with time-dependent analysis of the flow through LCS and is intended for the development of separation control strategies for aerodynamic surfaces.The compressible viscous flow over a NACA 65-(1)412 airfoil is solved with a DG based Navier-Stokes solver in two and three dimensions. A method is presented in which high-order polynomial element edges adjacent to curved boundaries are matched to boundaries defined by non-smooth splines. Artificial surface roughness introduced by the piecewise-linear boundary approximation of straight-sided meshes results in the simulation of incorrect physics, including wake instabilities and spurious time-dependent modes. Spectral accuracy in the boundary approximation is not achieved for non-analytic boundary functions, particularly in high curvature regions.An algorithm is developed for the high-order computation of Finite-Time Lyapunov Exponent (FTLE) fields simultaneously and efficiently with two and three dimensional DG-based flow solvers. Fluid tracers are initialized at Gauss-Lobatto quadrature nodes within an element and form the high-order basis for a flow map at later time. Gradients of the flow map and FTLE are evaluated with DG operators. Multiple flow maps are determined from a single particle trace by remapping the flow map to the quadrature nodes on deformed mesh elements. For large integration times, excessive subdomain deformation deteriorates the interpolating conditioning. The conditioning provides information on the fluid deformation and identifies subdomains that contain LCS. An exponential filter smooths the flow map in highly deformed areas. The algorithm is tested on several benchmarks and is shown to have spectral convergence.The two and three-dimensional LCS field are analyzed for the unsteady flow over a NACA 65-(1)412 airfoil at a free-stream Reynolds number of Re=20,000 based on the chord length and a Mach number of 0.3. In two-dimensions, a Karman vortex street forms at the trailing edge. The three-dimensional vortex street breaks down to turbulence at the trailing edge.

Background Some emergent large vessel occlusions (ELVOs) are refractory to reperfusion because of underlying intracranial atherosclerosis (ICAS), often requiring rescue therapy (RT) with balloon angioplasty, stenting, or both. In this study, we investigate the safety, efficacy, and long‐term outcomes of RT in the setting of mechanical thrombectomy for ICAS‐related ELVO. Methods and Results We queried the databases of 10 thrombectomy‐capable centers in North America and Europe included in STAR (Stroke Thrombectomy and Aneurysm Registry). Patients with ELVO who underwent ICAS‐related RT were included. A matched sample was produced for variables of age, admission National Institute of Health Stroke Scale, Alberta Stroke Program Early CT Score, onset to groin puncture time, occlusion site, and final recanalization. Out of 3025 patients with MT, 182 (6%) patients required RT because of underlying ICAS. Balloon angioplasty was performed on 122 patients, and 117 patients had intracranial stenting. In the matched analysis, 141 patients who received RT matched to a similar number of controls. The number of thrombectomy passes was higher (3 versus 1, P P =0.004). There was a higher rate of symptomatic hemorrhagic transformation in the RT group (7.8% versus 4.3%, P =0.211), however, the difference was not significant. There was no difference in 90‐day modified Rankin scale of 0 to 2 (44% versus 47.5%, P =0.543) between patients in the RT and control groups. Conclusions In patients with ELVO with underlying ICAS requiring RT, despite longer procedure time and a more thrombectomy passes, the 90 days favorable outcomes were comparable with patients with embolic ELVO.

Prostate cancer is the most highly diagnosed, non-cutaneous cancer and second leading cause of cancer-related death in the United States. Despite this, the molecular determinants that drive prostate cancer initiation and progression are still poorly understood. Identifying signaling networks that are activated in prostate cancer and how these pathways interact is therefore fundamental to our understanding of prostate cancer. Using a dissociated prostate tissue model, we are able to investigate the role of specific genes in the regeneration and transformation of prostate tissues. This system allows for interrogation of both cell-autonomous genes as well as paracrine factors secreted by the surrounding microenvironment. Further, prostate epithelial tumors can be obtained from defined oncogenic combinations and the pathways and interactions between these oncogenes interrogated through a variety of ex vivo techniques. We used the dissociated prostate tissue system to develop an array of transformation states using defined sets of oncogenes. These tumors were then interrogated using phosphotyrosine enrichment combined with mass spectrometry analysis to identify key signaling nodes activated by specific oncogenic combinations. By defining the signaling networks that are activated by specific oncogenic combinations, we begin to identify common signaling nodes that can be targeted therapeutically. Whole genome sequencing studies have shown that prostate cancer exhibits a relatively low mutation frequency. We hypothesized that oncogenic transformation could therefore be driven by the over-expression of non-mutated proteins that would then activate specific oncogenic pathways leading to transformation. We interrogated the effects of increased expression of Src kinase and the androgen receptor, and identified that upon heightened co-expression of the non-mutated forms was sufficient to drive prostate transformation. We then interrogated the role of inflammatory cytokines and their ability to drive prostate cancer initiation and progression using interleukin-6 (IL6) and the related oncostatin-M (OSM) cytokines. Increased expression of either IL6 or OSM was sufficient to drive progression of PTEN-initiated lesions, indicating functional synergy. Further, increased expression of these cytokines was associated with an increase in activation of pathways downstream of these inflammatory cytokines. These data indicate that heightened expression of non-mutated genes can promote activation of pathways associated with oncogenesis and are sufficient to drive prostate epithelial transformation.

I present results from several years of concerted observations of the afterglows and host galaxies of gamma-ray bursts (GRBs), the most energetic explosions in the Universe. Short gamma-ray bursts originate from a wide variety of environments, including disk galaxies, elliptical galaxies, galaxy haloes, and intracluster and intergalactic space. Long gamma ray bursts associate almost exclusively with star-forming hosts, but the properties of these galaxies also vary widely. Some are hosted in extremely small galaxies, difficult to identify directly in emission or infer from the absorption of afterglow light, but the host luminosity distribution extends up to very luminous (>L*) systems as well. A significant fraction of long GRBs are observed along highly dust-obscured sightlines through their host medium. Some of these events are hosted within conspicuously dusty galaxies, although the hosts of other dust-obscured events show no outward signs of significant internal dust content. By measuring the wavelength dependence of dust absorption profiles using a few well-observed GRB afterglows, I provide evidence for ordinary dust with properties similar to those of dust in the Milky Way in a system at z~3, but a very different absorption profile from the dust in a galaxy at z~5, providing tentative evidence to support a transition in dust composition early in the history of the Universe. I present an observationally-determined redshift distribution for Swift GRBs, showing few to originate from high redshifts (z>5). I also provide the first photometric and spectroscopic catalogs from one of the largest GRB host-galaxy surveys ever conducted, including observations of almost 150 distinct GRB fields.

Objective: We propose a computational framework for integrating diverse patient measurements into an aggregate health score and applying it to patient stability prediction. Materials and Methods: We mapped retrospective patient data from the Multiparameter Intelligent Monitoring in Intensive Care (MIMIC) II clinical database into a discrete multidimensional space, which was searched for measurement combinations and trends relevant to patient outcomes of interest. Patient trajectories through this space were then used to make outcome predictions. As a case study, we built AutoTriage, a patient stability prediction tool to be used for discharge recommendation. Results: AutoTriage correctly identified 3 times as many stabilizing patients as existing tools and achieved an accuracy of 92.9% (95% CI: 91.6–93.9%), while maintaining 94.5% specificity. Analysis of AutoTriage parameters revealed that interdependencies between risk factors comprised the majority of each patient stability score. Discussion: AutoTriage demonstrated an improvement in the sensitivity of existing stability prediction tools, while considering patient safety upon discharge. The relative contributions of risk factors indicated that time-series trends and measurement interdependencies are most important to stability prediction. Conclusion: Our results motivate the application of multidimensional analysis to other clinical problems and highlight the importance of risk factor trends and interdependencies in outcome prediction.

Estradiol (E2) administered to estrogen receptor-positive (ER(+)) breast cancer patients stimulates glucose uptake by tumors. Importantly, this E2-induced metabolic flare is predictive of the clinical effectiveness of anti-estrogens and, as a result, downstream metabolic regulators of E2 are expected to have utility as targets for the development of anti-breast cancer agents. The family of 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatases (PFKFB1-4) control glycolytic flux via their product, fructose-2,6-bisphosphate (F26BP), which activates 6-phosphofructo-1-kinase (PFK-1). We postulated that E2 might promote PFKFB3 expression, resulting in increased F26BP and glucose uptake. We demonstrate that PFKFB3 expression is highest in stage III lymph node metastases relative to normal breast tissues and that exposure of human MCF-7 breast cancer cells to E2 causes a rapid increase in [(14)C]glucose uptake and glycolysis that is coincident with an induction of PFKFB3 mRNA (via ER binding to its promoter), protein expression and the intracellular concentration of its product, F26BP. Importantly, selective inhibition of PFKFB3 expression and activity using siRNA or a PFKFB3 inhibitor markedly reduces the E2-mediated increase in F26BP, [(14)C]glucose uptake, and glycolysis. Furthermore, co-treatment of MCF-7 cells with the PFKFB3 inhibitor and the anti-estrogen ICI 182,780 synergistically induces apoptotic cell death. These findings demonstrate for the first time that the estrogen receptor directly promotes PFKFB3 mRNA transcription which, in turn, is required for the glucose metabolism and survival of breast cancer cells. Importantly, these results provide essential preclinical information that may allow for the ultimate design of combinatorial trials of PFKFB3 antagonists with anti-estrogen therapies in ER(+) stage IV breast cancer patients.

In human cancers, loss of PTEN, stabilization of hypoxia inducible factor-1α, and activation of Ras and AKT converge to increase the activity of a key regulator of glycolysis, 6-phosphofructo-2-kinase (PFKFB3). This enzyme synthesizes fructose 2,6-bisphosphate (F26BP), which is an activator of 6-phosphofructo-1-kinase, a key step of glycolysis. Previously, a weak competitive inhibitor of PFKFB3, 3-(3-pyridinyl)-1-(4-pyridinyl)-2-propen-1-one (3PO), was found to reduce the glucose metabolism and proliferation of cancer cells. We have synthesized 73 derivatives of 3PO and screened each compound for activity against recombinant PFKFB3. One small molecule, 1-(4-pyridinyl)-3-(2-quinolinyl)-2-propen-1-one (PFK15), was selected for further preclinical evaluation of its pharmacokinetic, antimetabolic, and antineoplastic properties in vitro and in vivo. We found that PFK15 causes a rapid induction of apoptosis in transformed cells, has adequate pharmacokinetic properties, suppresses the glucose uptake and growth of Lewis lung carcinomas in syngeneic mice, and yields antitumor effects in three human xenograft models of cancer in athymic mice that are comparable to U.S. Food and Drug Administration–approved chemotherapeutic agents. As a result of this study, a synthetic derivative and formulation of PFK15 has undergone investigational new drug (IND)-enabling toxicology and safety studies. A phase I clinical trial of its efficacy in advanced cancer patients will initiate in 2013 and we anticipate that this new class of antimetabolic agents will yield acceptable therapeutic indices and prove to be synergistic with agents that disrupt neoplastic signaling. Mol Cancer Ther; 12(8); 1461–70. ©2013 AACR.

ObjectiveTo develop high-performance early sepsis prediction technology for the general patient population. MethodsRetrospective analysis of adult patients admitted to the intensive care unit (from the MIMIC II dataset) who were not septic at the time of admission. ResultsA sepsis early warning algorithm, InSight, was developed and applied to the prediction of sepsis up to three hours prior to a patient's first five hour Systemic Inflammatory Response Syndrome (SIRS) episode. When applied to a never-before-seen set of test patients, InSight predictions demonstrated a sensitivity of 0.90 (95% CI: 0.89-0.91) and a specificity of 0.81 (95% CI: 0.80-0.82), exceeding or rivaling that of existing biomarker detection methods. Across predictive times up to three hours before a sustained SIRS event, InSight maintained an average area under the ROC curve of 0.83 (95% CI: 0.80-0.86). Analysis of patient sepsis risk showed that contributions from the coevolution of multiple risk factors were more important than the contributions from isolated individual risk factors when making predictions further in advance. ConclusionsSepsis can be predicted at least three hours in advance of onset of the first five hour SIRS episode, using only nine commonly available vital signs, with better performance than methods in standard practice today. High-order correlations of vital sign measurements are key to this prediction, which improves the likelihood of early identification of at-risk patients.

In this paper we present the design and implementation of the Linpack benchmark for the IBM BladeCenter QS22, which incorporates two IBM PowerXCell 8i1processors. The PowerXCell 8i is a new implementation of the Cell Broadband Engine™2 architecture and contains a set of special-purpose processing cores known as Synergistic Processing Elements (SPEs). The SPEs can be used as computational accelerators to augment the main PowerPC processor. The added computational capability of the SPEs results in a peak double precision floating point capability of 108.8 GFLOPS. We explain how we modified the standard open source implementation of Linpack to accelerate key computational kernels using the SPEs of the PowerXCell 8i processors. We describe in detail the implementation and performance of the computational kernels and also explain how we employed the SPEs for high-speed data movement and reformatting. The result of these modifications is a Linpack benchmark optimized for the IBM PowerXCell 8i processor that achieves 170.7 GFLOPS on a BladeCenter QS22 with 32 GB of DDR2 SDRAM memory. Our implementation of Linpack also supports clusters of QS22s, and was used to achieve a result of 11.1 TFLOPS on a cluster of 84 QS22 blades. We compare our results on a single BladeCenter QS22 with the base Linpack implementation without SPE acceleration to illustrate the benefits of our optimizations.

This paper describes the architecture and implementation of the original gaming-oriented synergistic processor element (SPE) in both 90-nm and 65-nm silicon-on-insulator (SOI) technology and introduces a new SPE implementation targeted for the high-performance computing community. The Cell Broadband Engine™ processor contains eight SPEs. The dual-issue, four-way single-instruction multiple-data processor is designed to achieve high performance per area and power and is optimized to process streaming data, simulate physical phenomena, and render objects digitally. Most aspects of data movement and instruction flow are controlled by software to improve the performance of the memory system and the core performance density. The SPE was designed as an 11-F04 (fan-out-of-4-inverter-delay) processor using 20.9 million transistors within 14.8 mm 2 using the IBM 90-nm SOI low-k process. CMOS (complementary metal-oxide semiconductor) static gates implement the majority of the logic. Dynamic circuits are used in critical areas and occupy 19% of the non-static random access memory (SRAM) area. Instruction set architecture, microarchitecture, and physical implementation are tightly coupled to achieve a compact and power-efficient design. Correct operation has been observed at up to 5.6 GHz and 7.3 GHz, respectively, in 90-nm and 65-nm SOI technology.

This paper provides an overview of the Cell Broadband Engine™ Architecture (CBEA). The CBEA defines a revolutionary extension to a more conventional processor organization and serves as the basis for the development of microprocessors targeted at the computer entertainment, multimedia, and real-time market segments. In this paper, the organization of the architecture is described, as well as the instruction set, commands, and facilities defined in the architecture. In many cases, the motivation for these facilities is explained and examples are provided to illustrate their intended use. In addition, this paper introduces the Software Development Kit and the software standards for a CBEA-compliant processor.

Purpose: PI3K-pathway activation is the second most common genetic abnormality occurring in head and neck squamous cell carcinoma (HNSCC). Mutation or amplification of PIK3CA, the gene coding for the p110α subunit of PI3K, occur in 34%-56% of HNSCC and there is evidence that targeting PI3Kα in these tumors can be radiosensitizing. Small molecule kinase inhibitors of PI3Kα are showing promising activity. However, the use of this molecule is limited by serious side effects such as hyperglycemia, diarrhea and rush. In this study we aimed to apply novel nanotechnology for targeted drug delivery via the cell adhesion molecule P-selectin, which was previously shown to be overexpressed in tumor microvasculature. Furthermore, P-selectin is up regulated following radiation therapy (RT), which could enhance drug delivery using targeted nanoparticles. Experimental design: We explored the efficacy and specificity of targeted delivery of BYL719, a potent alpha-specific PI3K inhibitor, using nanoparticles that selectively target P-selectin present in the tumor microvasculature in PIK3CA-mutated HNSCC. We compared the antitumor effects of nanoparticle delivery versus standard oral gavage with free drug in PIK3CA mutated HNSCC cell line- and patient-derived xenografts (PDXs). In addition, we compared blood glucose and insulin levels between standard BYL719 administration and nanoparticle drug delivery at different time points after treatment. Furthermore, we sought to evaluate the radiosensitizing properties of BYL719 when combined with fractionated RT of 5X4Gy. Results: P-selectin targeted delivery of 50 mg/kg/week BYL719 resulted in a significant tumor growth delay, which was comparable to the standard drug administration of 350 mg/kg/week (50 mg/kg/daily). While blood glucose and insulin levels were spiking after standard oral gavage of BYL719, these parameters were virtually unchanged upon nanoparticle administration of the drug. The radiosensitizing abilities of low-dose nanoparticle-linked BYL719 were comparable to the drug concentrations used in standard daily BYL719 administration. When given concomitantly with fractionated RT, both methods showed robust enhancement of radiotherapy response in all the models tested resulting in durable control of tumor growth. Conclusions: P-selectin targeted delivery of a PI3Kα inhibitor resulted in inhibition of the PI3K/Akt/mTOR pathway without affecting glucose or insulin levels. The magnitude of pathway inhibition was sufficient to radiosensitize several HNSCC animal models. This novel targeting strategy could be translated to the clinic to treat patients with PIK3CA activated and radioresistant HNSCC tumors sparing most of the systemic adverse effects of PI3K inhibition. Citation Format: Aviram Mizrachi, Yosi Shamay, Janki Shah, Samuel I. Brook, Joanne Soong, Vinagolu K. Rajasekhar, John L. Humm, Simon N. Powell, José Baselga, Daniel A. Heller, Adriana Haimovitz-Friedman, Maurizio Scaltriti. Tumor-specific PI3K inhibition via nanoparticle targeted delivery in head and neck squamous cell carcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 4122. doi:10.1158/1538-7445.AM2017-4122

It is controversial whether acute myeloid leukemia (AML) patients with 20-29% bone marrow (BM) blasts, formerly referred to as refractory anemia with excess blasts in transformation (RAEBT), should be considered AML or myelodysplastic syndrome (MDS) for the purposes of treatment and prognostication. We retrospectively studied 571 de novo AML in patients aged50 years, including 142 RAEBT and 429 with ≥30% blasts (AML30), as well as 151 patients with 10-19% BM blasts (RAEB2). RAEBT patients were older and had lower white blood count, but higher hemoglobin, platelet count, and karyotype risk scores compared to AML30, while these features were similar to RAEB2. FLT3 and NPM1 mutations and monocytic morphology occurred more commonly in AML30 than in RAEBT. RAEBT patients were treated less often with induction therapy than AML30, whereas allogeneic stem cell transplant frequency was similar. The median and 4-year OS of RAEBT patients were longer than those of AML30 patients (20.5 vs 12.0 months and 28.6% vs 20.4%, respectively, P = 0.003); this difference in OS was manifested in patients in the intermediate UKMRC karyotype risk group, whereas OS of RAEBT patients and AML30 patients in the adverse karyotype risk group were not significantly different. Multivariable analysis showed that RAEBT (P 0.0001), hemoglobin (P = 0.005), UKMRC karyotype risk group (P = 0.002), normal BM karyotype (P = 0.004), treatment with induction therapy (P 0.0001), and stem cell transplant (P 0.0001) were associated with longer OS. Our findings favor considering de novo RAEBT as a favorable prognostic subgroup of AML.

This study was undertaken to determine the effects of high fat diet and aldose reductase deficiency on the structure and function of murine perivascular adipose tissue. Wild type (WT) (C57 bl/6) and aldose reductase knockout (AR-null) mice were aged to 8 or 18 weeks on a normal chow diet then some were switched to a high fat diet for 4 weeks while others remained on the control diet for the same amount of time. Following 4 weeks, mice were euthanized PVAT was analyzed. High fat feeding significantly increased body fat percentage in both age groups of mice and the area of PVAT present around distal thoracic aorta sections also increased. Additionally, WT and AR-null mice appeared to exhibit an increase in the area of white adipocyte clusters present in PVAT, although the results were not significant. When examining the area of white adipocytes present in PVAT, HF feeding significantly increased area of adipocytes throughout PVAT in WT, 22-week old mice and a significant increase in VAT and overall PVAT, but not DAT in AR-null mice. To examine the functional properties of PVAT, sections of aorta with intact and removed PVAT were measured for contraction and relaxation properties. In WT mice, HF feeding resulted in endothelial dysfunction, seen as a reduced relaxation response to acetylcholine. The presence of intact PVAT reversed the observed dysfunction. Collectively, these data suggest that PVAT adiposity is increased similar to overall adiposity with high fat feeding and that this change is structure may contribute to the functional changes also observed with high fat feeding.

Purpose: Patients with primary breast carcinomas expressing of both estrogen (ERα) and progesterone receptors (PR) are more likely to respond to Tamoxifen therapy than those with ERα and PR negative cancers, especially in women with early stage lesions. However, a significant number of these patients exhibiting clinicopathologic features suggesting good response and prognosis will relapse within 10 years. Additional markers are needed to identify patients at risk for breast cancer recurrence after administrative hormone therapy. Procedures: Ten candidate genes associated with estrogen signaling were selected from a group of 200 genes identified from microarray studies of laser capture microdissected breast carcinoma cells of ER positive breast cancers and combined with those for conventional biomarkers (ESR1, PGR, ERBB2). Carcinoma cell content and pathology of each biopsy was confirmed. Expression of this 13-gene subset was analyzed by RT-qPCR in de-identified frozen tissue specimens from early stage, ER+/PR+ breast cancers treated with adjuvant Tamoxifen. A multivariate model was created by Cox regression using a training set of tissue sections from 36 biopsy specimens, and then applied to an independent validation set of sections from 24 specimens. Results: A 5-gene model was derived from the training set that exhibited significant correlations with both relapse-free and overall survival. Applying this model to Kaplan-Meier regression, patients were separated into low risk (100% relapse-free at 150 months) and high risk (60% relapse-free at 150 months) groups (P = 0.03). When this model was applied to the validation set, similar risk stratification was achieved for both relapse-free and overall survival (P = 0.01 and 0.04, respectively). No significant difference in clinical features (e.g., tumor grade, tumor size and nodal status) was observed between the high and low risk groups. Conclusions: We developed a 5-gene model composed of PgR, BCL2, ERBB4 JM-a, RERG and CD34 that identified a subset of early stage, ER+/PR+ breast cancers patients treated with Tamoxifen that exhibited high recurrence rates, although their clinicopathologic features suggested good prognosis and response. Furthermore, absence of association of the 5-gene model with tumor grade suggests the gene expression profile provides clinically relevant predictive information independent of proliferation. Supported in part by grants from the Phi Beta Psi Sorority Charity Trust and a CTSP award from the Commonwealth of Kentucky. DAK was a recipient of a Graduate Fellowship from the Integrated Programs in Biomedical Sciences, University of Louisville, and the Spatola Graduate Fellowship from IMD3. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 4413. doi:10.1158/1538-7445.AM2011-4413

We describe the challenges and opportunities we encountered when developing a hybrid version of the Linpack benchmark for the Los Alamos National Laboratory Roadrunner supercomputing system, which combines traditional x86-64 host processors with IBM PowerXCell™ 8i accelerator processors. The challenges included determining the proper division of the host and accelerator roles in the computation, transfer of data between the host and accelerator memory domains, alignment of data for communication and computation, and data format differences between the two processors. We also describe our approach to modeling the performance of the hybrid system and compare our performance estimates to witnessed performance on the system at different scales and levels of memory consumption. Through careful attention to these issues, we have produced a hybrid version of the Linpack benchmark for the Roadrunner system that achieves 77.8% of peak performance on a single compute node and 74.6% of peak performance over the entire system, making this system the first to achieve a Linpack result exceeding one petaflops (1015 floating-point operations per second).