Publications

Speedup, communication complexity and blocking-a La Recherche du Temps Perdu

The paper investigates the time lost in a parallel computation due to sequential and duplicated work, communication and control, and blocking. It introduces the concept of relative speedup and proposes characterizations of parallel algorithms based upon the communication complexity and the blocking model. The paper discusses the impact of the processor's architecture upon the measured speedup. It shows that a large speedup may be due to an inefficient sequential computation, e.g. due to the cache management, rather than to an efficient parallel computation. A model of parallel computations which takes into account sequential and duplicated work, communication and control and blocking is presented. The paper shows that the scalability of a parallel computation is determined by the communication complexity. The model is used to predict the asymptotic behavior, the maximum speedup and the optimal number of processors. An incore 3D FFT algorithm for distributed memory MIMD systems and a Chebyshev iterative algorithm for solving a linear system of equations are used to illustrate the concepts.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

Stability of the Associations between Early Life Risk Indicators and Adolescent Overweight over the Evolving Obesity Epidemic

BACKGROUND Pre- and perinatal factors and preschool body size may help identify children developing overweight, but these factors might have changed during the development of the obesity epidemic. OBJECTIVE We aimed to assess the associations between early life risk indicators and overweight at the age of 9 and 15 years at different stages of the obesity epidemic. METHODS We used two population-based Northern Finland Birth Cohorts including 4111 children born in 1966 (NFBC1966) and 5414 children born in 1985-1986 (NFBC1986). In both cohorts, we used the same a priori defined prenatal factors, maternal body mass index (BMI), birth weight, infant weight (age 5 months and 1 year), and preschool BMI (age 2-5 years). We used internal references in early childhood to define percentiles of body size ( 90) and generalized linear models to study the association with overweight, according to the International Obesity Taskforce (IOTF) definitions, at the ages of 9 and 15 years. RESULTS The prevalence of overweight at the age of 15 was 9% for children born in 1966 and 16% for children born in 1986. However, medians of infant weight and preschool BMI changed little between the cohorts, and we found similar associations between maternal BMI, infant weight, preschool BMI, and later overweight in the two cohorts. At 5 years, children above the 90th percentile had approximately a 12 times higher risk of being overweight at the age of 15 years compared to children below the 50th percentile in both cohorts. CONCLUSIONS The associations between early body size and adolescent overweight showed remarkable stability, despite the increase in prevalence of overweight over the 20 years between the cohorts. Using consequently defined internal percentiles may be a valuable tool in clinical practice.

Early life factors and blood lipid concentrations at age 31 in the 1966 northern Finland birth cohort

Multiple Output Regression with Latent Noise

In high-dimensional data, structured noise caused by observed and unobserved factors affecting multiple target variables simultaneously, imposes a serious challenge for modeling, by masking the often weak signal. Therefore, (1) explaining away the structured noise in multiple-output regression is of paramount importance. Additionally, (2) assumptions about the correlation structure of the regression weights are needed. We note that both can be formulated in a natural way in a latent variable model, in which both the interesting signal and the noise are mediated through the same latent factors. Under this assumption, the signal model then borrows strength from the noise model by encouraging similar effects on correlated targets. We introduce a hyperparameter for the latent signal-to-noise ratio which turns out to be important for modelling weak signals, and an ordered infinite-dimensional shrinkage prior that resolves the rotational unidentifiability in reduced-rank regression models. Simulations and prediction experiments with metabolite, gene expression, FMRI measurement, and macroeconomic time series data show that our model equals or exceeds the state-of-the-art performance and, in particular, outperforms the standard approach of assuming independent noise and signal models.

Whole Grain, Bran, and Germ Intake and Risk of Type 2 Diabetes: A Prospective Cohort Study and Systematic Review

Whole grain intake is inversely associated with risk of type 2 diabetes, and this association is stronger for bran than for germ. Findings from prospective cohort studies consistently support increasing whole grain consumption for the prevention of type 2 diabetes.

Maternal Glycemic Dysregulation During Pregnancy and Neonatal Blood DNA Methylation: Meta-analyses of Epigenome-Wide Association Studies

Maternal glycemic dysregulation during pregnancy increases the risk of adverse health outcomes in her offspring, a risk thought to be linearly related to maternal hyperglycemia. It is hypothesized that changes in offspring DNA methylation (DNAm) underline these associations. To address this hypothesis, we conducted fixed-effects meta-analyses of epigenome-wide association study (EWAS) results from eight birth cohorts investigating relationships between cord blood DNAm and fetal exposure to maternal glucose (Nmaximum= 3,503), insulin (Nmaximum= 2,062), and area under the curve of glucose (AUCgluc) following oral glucose tolerance tests (Nmaximum = 1,505). Inflation and bias was addressed with the bacon Bioconductor package for individual cohorts and the meta-analysis results and reported model estimates and standard errors are adjusted. Greater maternal AUCgluc during pregnancy was associated with lower cord blood DNAm at neighboring CpGs cg26974062 and cg02988288 in TXNIP. Follow-up analyses show that these changes may have functional and later-life consequences and warrant further mediation and causal analysis. The corresponding paper can be found here: https://doi.org/10.2337/dc21-1701. Upload of this dataset was completed by The EWAS Catalog team. The data can be queried along with hundreds of other EWAS at ewascatalog.org. To upload your EWAS summary statistics and have a Zenodo DOI generated for you go to ewascatalog.org/upload.

Lipid Profiles and Heart Failure Risk

Rationale: Altered lipid metabolism has been implicated in heart failure (HF) development, but no prospective studies have examined comprehensive lipidomics data and subsequent risk of HF. Objective: We aimed to link single lipid metabolites and lipidomics networks to the risk of developing HF. Methods and Results: Discovery analyses were based on 216 targeted lipids in a case-control study (331 incident HF cases and 507 controls, matched by age, sex, and study center), nested within the PREDIMED (Prevención con Dieta Mediterránea) study. Associations of single lipids were examined in conditional logistic regression models. Furthermore, lipidomics networks were linked to HF risk in a multistep workflow, including machine learning–based identification of the HF-related network clusters, and regression-based discovery of the HF-related lipid patterns within these clusters. If available, significant findings were externally validated in a subsample of the EPIC-Potsdam cohort (2414 at-risk participants, including 87 incident HF cases). After confounder-adjustments, 2 lipids were significantly associated with HF risk in both cohorts: CER (ceramide) 16:0 (relative risk [RR] per SD in PREDIMED, 1.28 [95% CI, 1.13–1.47]) and phosphatidylcholine 32_0 (RR per SD in PREDIMED, 1.23 [95% CI, 1.08–1.41]). Additionally, lipid patterns in several network clusters were associated with HF risk in PREDIMED. Adjusted for standard risk factors, an internally cross-validated score based on the significant HF-related lipids that were identified in the network analysis in PREDIMED was associated with a higher HF risk (20 lipids, RR per SD, 2.33 [95% CI, 1.93%–2.81%). Moreover, a lipid score restricted to the externally available lipids was significantly associated with HF incidence in both cohorts (6 lipids, RRs per SD, 1.30 [95% CI, 1.14–1.47] in PREDIMED, and 1.46 [95% CI, 1.17–1.82] in EPIC-Potsdam). Conclusions: Our study identified and validated 2 lipid metabolites and several lipidomics patterns as potential novel biomarkers of HF risk. Lipid profiling may capture preclinical molecular alterations that predispose for incident HF. Registration: URL: https://www.isrctn.com/ISRCTN35739639 ; Unique identifier: ISRCTN35739639.

Choline Metabolism and Risk of Atrial Fibrillation and Heart Failure in the PREDIMED Study

Abstract Background Few studies have examined the associations of trimethylamine-N-oxide (TMAO) and its precursors (choline, betaine, dimethylglycine, and L-carnitine) with the risk of atrial fibrillation (AF) and heart failure (HF). This study sought to investigate these associations. Methods Prospective associations of these metabolites with incident AF and HF were examined among participants at high cardiovascular risk in the PREDIMED study (PREvención con DIeta MEDiterránea) after follow-up for about 10 years. Two nested case-control studies were conducted, including 509 AF incident cases matched to 618 controls and 326 HF incident cases matched to 426 controls. Plasma levels of TMAO and its precursors were semi-quantitatively profiled with liquid chromatography tandem mass spectrometry. Odds ratios were estimated with multivariable conditional logistic regression models. Results After adjustment for classical risk factors and accounting for multiple testing, participants in the highest quartile vs. the lowest quartile of baseline choline and betaine levels had a higher risk of AF [OR (95% CI): 1.85 (1.30–2.63) and 1.57 (1.09–2.24), respectively]. The corresponding OR for AF for extreme quartiles of dimethylglycine was 1.39 (0.99–1.96). One SD increase in log-transformed dimethylglycine was positively associated with AF risk (OR, 1.17; 1.03–1.33). The corresponding ORs for HF for extreme quartiles of choline, betaine, and dimethylglycine were 2.51 (1.57–4.03), 1.65 (1.00–2.71) and 1.65 (1.04–2.61), respectively. TMAO and L-carnitine levels were not associated with AF or HF. Conclusions Our findings support the role of the choline metabolic pathway in the pathogenesis of AF and HF.

Common variants near MC4R are associated with fat mass, weight and risk of obesity

Abstract 10: Increased Consumption of Fruits and Vegetables and Weight Change Over Time

Introduction: Current dietary guidelines recommend eating a variety of fruits and vegetables. However, based on nutrient composition, specific fruits and vegetables may be more or less beneficial for weight maintenance. Hypothesis: Fruits and vegetables with a higher fiber content or lower glycemic load are more strongly associated with weight maintenance and even weight loss, in part through mechanisms related to increased satiety. Methods: We examined the association between increased consumption of fruits and vegetables and change in body weight among participants &lt;65 years of age from three large prospective cohort studies: 41,608 women in the Nurses’ Health Study (NHS) I, 20,143 men in the Health Professionals Follow-up Study, and 67,533 women in the NHS II. Beginning in 1986, these associations were examined during 24 years of follow-up within 4-year time intervals, adjusting for simultaneous changes in other lifestyle factors such as diet, smoking, and physical activity. Results: Increased consumption of one serving per day of several types of fruits was inversely associated with weight gain over 4 years, for example -1.11 lbs (95% CI, -1.43 to -0.80 lbs) for berries and -0.96 lbs (95% CI, -1.20 to -0.72 lbs) for apples/pears. Increased consumption of several vegetables was also inversely associated with weight gain, for example -2.15 lbs (95% CI, -2.58 to -1.73 lbs) for tofu/soy and -1.18 lbs (95% CI, -1.73 to -0.64 lbs) for cauliflower. On the other hand, increased consumption of some vegetables was positively associated with weight change, for example corn (1.69 lbs; 95% CI, 0.90 to 2.48 lbs) and peas (0.81 lbs; 95% CI 0.16 to 1.46 lbs). The magnitude of weight change associated with increased consumption of high fiber vegetables (-0.37 lbs; 95% CI -0.48 to -0.27 lbs) was larger than that for low fiber vegetables (-0.13 lbs; 95% CI -0.23 to -0.02 lbs). Increased consumption of low glycemic load vegetables was associated with a greater magnitude of weight change, -0.37 lbs (95% CI, -0.63 to -0.11 lbs), compared to high glycemic load vegetables, -0.01 lbs (95% CI, -0.14 to 0.11 lbs). In contrast, weight change associated with fruit consumption did not vary by fiber content or glycemic load. Conclusions: Increased consumption of total fruits and vegetables was inversely associated with weight change over 4 years, however we found wide-ranging associations for specific foods, partly due to the fiber content and glycemic load of vegetables but not fruits.

Mining the performance of complex systems

The enormous amount of data that can be collected in any performance evaluation study of a complex system indicates the need for methodologies and systems capable of analyzing, fusing and reducing high-dimensional data spaces with very high speed. In this paper, we devise and present an adaptation of the Knowledge Discovery in Databases (KDD) framework introduced by U. Fayyad et al. (1996) that supports the above functionality for performance data of complex software/hardware system pairs. The KDD framework considered integrates database technology along with data mining techniques for uncovering patterns from performance data and static system characteristics. A case study is presented to demonstrate the effectiveness and applicability of the KDD approach for the performance evaluation of complex systems. The data mining tools utilized are general-purpose, public-domain and independent of the specific performance database involved. We are currently implementing the proposed KDD framework within an end-to-end performance evaluation system for designing complex parallel and distributed systems referred to as POEMS (Performance-Oriented End-to-end Modeling System).

Understanding the genetic complexity of puberty timing across the allele frequency spectrum

Abstract Pubertal timing varies considerably and has been associated with a range of health outcomes in later life. To elucidate the underlying biological mechanisms, we performed multi-ancestry genetic analyses in ∼800,000 women, identifying 1,080 independent signals associated with age at menarche. Collectively these loci explained 11% of the trait variance in an independent sample, with women at the top and bottom 1% of polygenic risk exhibiting a ∼11 and ∼14-fold higher risk of delayed and precocious pubertal development, respectively. These common variant analyses were supported by exome sequence analysis of ∼220,000 women, identifying several genes, including rare loss of function variants in ZNF483 which abolished the impact of polygenic risk. Next, we implicated 660 genes in pubertal development using a combination of in silico variant-to-gene mapping approaches and integration with dynamic gene expression data from mouse embryonic GnRH neurons. This included an uncharacterized G-protein coupled receptor GPR83 , which we demonstrate amplifies signaling of MC3R , a key sensor of nutritional status. Finally, we identified several genes, including ovary-expressed genes involved in DNA damage response that co-localize with signals associated with menopause timing, leading us to hypothesize that the ovarian reserve might signal centrally to trigger puberty. Collectively these findings extend our understanding of the biological complexity of puberty timing and highlight body size dependent and independent mechanisms that potentially link reproductive timing to later life disease.

The enzymatic synthesis of 1,5-dideoxy-1,5-diimino-d-talitol and 1-deoxygalactostatin using fuculose-1-phosphate aldolase

P116 Developing a risk prediction tool for spirometrically defined COPD in adults; experience from the ageing lungs in european cohorts project

<h3>Background</h3> The EU H2020 funded Ageing Lungs in European Cohorts study (ALEC www.alecstudy.org) is a consortia of population-based birth and adult inception cohorts with repeated measures of spirometry several years apart. Using these data, ALEC aims to identify risk factors for spirometrically defined COPD, and to develop an online risk prediction tool for use by the public/physicians. Our systematic review of available risk prediction tools (Matheson <i>et al</i> IJCOPD 2018;13:1927–35) showed previous attempts had used numerous variables (most commonly smoking, sex and age), but current models could not accurately rule in nor rule out future risk of COPD. <h3>Method</h3> Using Bayesian statistical methods we developed a model to predict a subject's lifetime risk of developing COPD (FEV1/FVC&lt;LLN). This model can use information from all cohorts even when some predictors of interest have not been collected within a cohort. Compared with existing approaches to deal with variables missing across studies, our Bayesian approach performs well and is flexible. Furthermore, it allows inclusion in the model of evidence from elsewhere on the effect of specific predictors if available and appropriate. Using R Shiny, we developed an online prototype interface that implements our prediction model allowing users to answer questions on their physical characteristics and lifestyle to generate a personalised risk of COPD. This prototype online prediction tool was reviewed by a group of researchers and clinicians. <h3>Results</h3> R shiny provides a technically suitable web-based interface for input of personal information and linkage to model outputs. Feedback from potential users was cautiously positive with concerns expressed regarding presentation of risk, provision of information to patients directing them for further health advice, limitation of underlying data to white Caucasian populations, interpretation/management of 'low risk of COPD' in those who smoked, and absence of risk prediction of exacerbations in those with established disease. <h3>Conclusion</h3> Using appropriate statistical methods, it is possible to develop a risk prediction model by combining data across cohorts even when they have not all collected exactly the same information. R Shiny provides a user-friendly means to create online disease risk prediction tools; however many challenges remain regarding full-scale implementation. These are common to many risk prediction tools.

Understanding lipidomics associations and the lipoprotein-related caveats in population epidemiology

Abstract Mass spectrometry lipidomics is becoming customary to analyse serum/plasma samples in epidemiology. The measurables are molecular constituents of lipoprotein particles, but very little is known on the consequences of adjusting lipidomics data with lipoprotein measures. We studied two population cohorts with 5,657 and 2,036 participants. LC-MS/MS lipidomics was applied to analyse 24 molecular lipid classes and NMR spectroscopy to quantify seven lipoprotein lipids plus apolipoprotein A-I (apoA-I) and B (apoB). The associations of these measures were analysed via partial Spearman’s correlations. The effects of nine different lipoprotein adjustments on these interrelationships were assessed. Multivariable regression modelling with these adjustments was also performed for the associations between the lipidomics data and BMI. These novel large-scale lipidomics data and their associations between the lipoprotein measures were coherent in both population cohorts, confirming the compatibility of the analytical approaches. Simulated data were generated to corroborate the mediation effects. The lipoprotein-related lipid-transport and metabolism inherently mediate the lipidomics associations as evident from the striking effects of the lipoprotein adjustments. These effects and their relevance to the interpretations of lipidomics data are presented and discussed in detail for the first time. The combined lipoprotein lipid adjustments appear prone to overadjustment and arbitrary biases.

Stitching Flexible Electronics into the Brain

Understanding complex neuronal networks requires monitoring long‐term neuronal activity in various regions of the brain. Significant progress has been made in multisite implantations of well‐designed probes, such as multisite implantation of Si‐based and polymer‐based probes. However, these multiprobe strategies are limited by the sizes and weights of interfaces to the multiple probes and the inability to track the activity of the same neurons and changes in neuronal activity over longer time periods. Here, a long single flexible probe that can be implanted by stitching into multiple regions of the mouse brain and subsequently transmit chronically stable neuronal signals from the multiple sites via a single low‐mass interface is reported. The probe at four different sites is implemented using a glass capillary needle or two sites using an ultrathin metal needle. In vitro tests in brain‐mimicking hydrogel show that multisite probe implantations achieve a high connection yield of &gt;86%. In vivo histological images at each site of probes, implanted by stitching using either glass capillary or ultrathin metal insertion needles exhibit seamless tissue–probe interfaces with negligible chronic immune response. In addition, electrophysiology studies demonstrate the ability to track single neuron activities at every injection site with chronic stability over at least one month. Notably, the measured spike amplitudes and signal‐to‐noise ratios at different implantation sites show no statistically significant differences. Multisite stitching implantation of flexible electronics in the brain opens up new opportunities for both fundamental neuroscience research and electrotherapeutic applications.

Psychosocial Predictors of Different Stages of Cigarette Smoking among High School Students

Scanning Tunneling Microscopy Studies of Low-Dimensional Materials: Charge Density Wave Pinning and Melting in Two Dimensions

Machine learning (ML) is transforming all areas of science. The complex and time-consuming calculations in molecular simulations are particularly suitable for an ML revolution and have already been profoundly affected by the application of existing ML ...Read More

Regeneration of nicotinamide cofactors for use in organic synthesis