Effect of training-sample size and classification difficulty on the accuracy of genomic predictors.

Introduction: As part of the MicroArray Quality Control (MAQC)-II project, this analysis examines how the choice of univariate feature-selection methods and classification algorithms may influence the performance of genomic predictors under varying degrees of prediction difficulty represented by three clinically relevant endpoints. Methods: We used gene-expression data from 230 breast cancers (grouped into training and independent validation sets), and we examined 40 predictors (five univariate feature-selection methods combined with eight different classifiers) for each of the three endpoints. Their classification performance was estimated on the training set by using two different resampling methods and compared with the accuracy observed in the independent validation set. Results: A ranking of the three classification problems was obtained, and the performance of 120 models was estimated and assessed on an independent validation set. The bootstrapping estimates were closer to the validation performance than were the cross-validation estimates. The required sample size for each endpoint was estimated, and both gene-level and pathway-level analyses were performed on the obtained models. Conclusions: We showed that genomic predictor accuracy is determined largely by an interplay between sample size and classification difficulty. Variations on univariate feature-selection methods and choice of classification algorithm have only a modest impact on predictor performance, and several statistically equally good predictors can be developed for any given classification problem.

Male dominance linked to size and age, but not to 'good genes' in brown trout (Salmo trutta).

BACKGROUND: Males that are successful in intra-sexual competition are often assumed to be of superior quality. In the mating system of most salmonid species, intensive dominance fights are common and the winners monopolise most mates and sire most offspring. We drew a random sample of mature male brown trout (Salmo trutta) from two wild populations and determined their dominance hierarchy or traits linked to dominance. The fish were then stripped and their sperm was used for in vitro fertilisations in two full-factorial breeding designs. We recorded embryo viability until hatching in both experiments, and juvenile survival during 20 months after release into a natural streamlet in the second experiment. Since offspring of brown trout get only genes from their fathers, we used offspring survival as a quality measure to test (i) whether males differ in their genetic quality, and if so, (ii) whether dominance or traits linked to dominance reveal 'good genes'. RESULTS: We found significant additive genetic variance on embryo survival, i.e. males differed in their genetic quality. Older, heavier and larger males were more successful in intra-sexual selection. However, neither dominance nor dominance indicators like body length, weight or age were significantly linked to genetic quality measured as embryo or juvenile survival. CONCLUSION: We found no evidence that females can improve their offspring's genetic viability by mating with large and dominant males. If there still were advantages of mating with dominant males, they may be linked to non-genetic benefits or to genetic advantages that are context dependent and therefore possibly not revealed under our experimental conditions - even if we found significant additive genetic variation for embryo viability under such conditions.

Une société de classes moyennes ? Le cas de la société suisse contemporaine

Depuis les années 1980, la sociologie a connu une résurgence des thèses du déclin des classes sociales qui a provoqué d'importants débats. En matière d'évolution de la structure de classe plus particulièrement, le débat se focalise sur les thèses dites de la « moyennisation » de la société et de l'« aspiration vers le haut ». Dans ce cadre, la société (postindustrielle) serait désormais constituée d'une vaste classe moyenne tendant à provoquer une dissolution générale des frontières de classes. Cet article est une contribution empirique à ce débat et porte, plus précisément, sur l'évolution de la société suisse contemporaine. A priori, le cas suisse peut sembler favorable aux thèses du déclin des classes sociales. En effet, en comparaison européenne, la Suisse connaît un secteur des services supérieur à la moyenne, un chômage inférieur à celle-ci ou encore un niveau de rémunération supérieur à la moyenne ; tous phénomènes souvent mis en avant pour expliquer le déclin affirmé des classes sociales. Afin d'évaluer les thèses sous revue, une série d'hypothèses ont été formulées au sujet de l'évolution de la division du travail (mesurée selon la classification internationale des professions) d'une part, de la structure de classe (approchée par les catégories socioprofessionnelles helvétiques) d'autre part. L'analyse se base sur des données secondaires, ce qui implique une dépendance vis-à-vis des pratiques des institutions « productrices ». Malgré ces limitations méthodologiques, l'analyse montre que - au moins prises comme un tout cohérent - les thèses du déclin des classes sociales sont réfutées dans le cas, pourtant favorable en principe, de la société suisse contemporaine. Ainsi, entre autres, cette dernière s'avère n'être pas plus une société de classes moyennes que de classes populaires ; ce qui la rapproche d'ailleurs de nombreuses sociétés européennes.

Adult maternal body size matters.

Comment on: Horta BL, Gigante DP, Osmond C, Barros FC, Victora CG. Intergenerational effect of weight gain in childhood on offspring birthweight. Int J Epidemiol. 2009 Jun;38(3):724-32. PMID: 19376883.

Aggregate cultures of foetal rat liver cells: development and maintenance of liver gene expression.

Rotation-mediated aggregate cultures of foetal rat liver cells were prepared and grown in a chemically defined medium. Their capacity for cellular organisation and maturation was studied over a culture period of 3 wk by using both morphologic and biochemical criteria. It was found that within each aggregate, distinct liver cell types were present and attained their normal, differentiated phenotype. Parenchymal cells formed small acini with a central lumen. Within the first 2 wk in culture, albumin and ferritin mRNA levels were maintained, while the alpha-fetoprotein mRNA levels decreased, and tyrosine aminotransferase (TAT) gene expression increased. No significant response to glucocorticoids was observed in early cultures, whereas after 3 wk a marked increase in TAT mRNA levels was elicited by dexamethasone and glucagon (additive stimulatory effects). The results show that foetal rat liver cells cultured in a chemically defined medium are able to rearrange themselves into histotypic structures, and display a developmental pattern of gene expression comparable to that of perinatal rat liver in vivo. This culture system offers therefore a useful model to study the development and function of liver cells.

Dynamical modeling and analysis of large cellular regulatory networks.

The dynamical analysis of large biological regulatory networks requires the development of scalable methods for mathematical modeling. Following the approach initially introduced by Thomas, we formalize the interactions between the components of a network in terms of discrete variables, functions, and parameters. Model simulations result in directed graphs, called state transition graphs. We are particularly interested in reachability properties and asymptotic behaviors, which correspond to terminal strongly connected components (or "attractors") in the state transition graph. A well-known problem is the exponential increase of the size of state transition graphs with the number of network components, in particular when using the biologically realistic asynchronous updating assumption. To address this problem, we have developed several complementary methods enabling the analysis of the behavior of large and complex logical models: (i) the definition of transition priority classes to simplify the dynamics; (ii) a model reduction method preserving essential dynamical properties, (iii) a novel algorithm to compact state transition graphs and directly generate compressed representations, emphasizing relevant transient and asymptotic dynamical properties. The power of an approach combining these different methods is demonstrated by applying them to a recent multilevel logical model for the network controlling CD4+ T helper cell response to antigen presentation and to a dozen cytokines. This model accounts for the differentiation of canonical Th1 and Th2 lymphocytes, as well as of inflammatory Th17 and regulatory T cells, along with many hybrid subtypes. All these methods have been implemented into the software GINsim, which enables the definition, the analysis, and the simulation of logical regulatory graphs.

DETERMINATION OF CELL-SPECIFIC NEUROTOXICITY OF MALONATE, METHYLMALONATE AND PROPIONATE IN A 3D RAT BRAIN CELL AGGREGATE SYSTEM

Malonate, methylmalonate and propionate are potentially neurotoxic metabolites in branched-chain organic acidurias. Their effects were tested on cultured 3D rat brain cell aggregates, using dosages of 0.1, 1.0 and 10.0 mM with a short but intense (twice a day over 3 days) and a longer but less intense treatment (every 3 rdday over 9 days). CNS cell-specific immunohistochemical stainings allowed the follow-up of neurons (axons, phosphorylated medium-weight neurofilament), astrocytes (glial fibrillary acidic protein) and oligodendrocytes (myelin basic protein). Methylmalonate and malonate were quantified by tandem mass spectrometry. Tandem mass spectrometry analysis of harvested brain cell aggregates revealed clear intracellular accumulation of methylmalonate and malonate. In immunohistochemical stainings oligodendrocytes appeared the most affected brain cells. The MBP signal disappeared already at 0.1 mM treatment with each metabolite. Mature astrocytes were not affected by propionate, while immature astrocytes on intense treatment with propionate developed cell swelling. 1 mM methylmalonate induced cell swelling of both immature and mature astrocytes , while 1 mM malonate only affected mature astrocytes. Neurons were not affected by methylmalonate, but 10.0 mM malonate on less intense treatment and 0.1, 1.0 and 10.0 mM propionate on intense treatment affected axonal growth. Our study shows significant uptake and deleterious effects of these metabolites on brain cells, principally on astrocytes and oligodendrocytes. This may be explained by the absence of the pathway in glial cells, which thus are not able to degrade these metabolites. Further studies are ongoing to elucidate the underlying mechanisms of the observed neurotoxic effects.

Relating the permeability of quartz sands to their grain size and spectral induced polarization characteristics

Recently, Revil & Florsch proposed a novel mechanistic model based on the polarization of the Stern layer relating the permeability of granular media to their spectral induced polarization (SIP) characteristics based on the formation of polarized cells around individual grains. To explore the practical validity of this model, we compare it to pertinent laboratory measurements on samples of quartz sands with a wide range of granulometric characteristics. In particular, we measure the hydraulic and SIP characteristics of all samples both in their loose, non-compacted and compacted states, which might allow for the detection of polarization processes that are independent of the grain size. We first verify the underlying grain size/permeability relationship upon which the model of Revil & Florsch is based and then proceed to compare the observed and predicted permeability values for our samples by substituting the grain size characteristics by corresponding SIP parameters, notably the so-called Cole-Cole time constant. In doing so, we also asses the quantitative impact of an observed shift in the Cole-Cole time constant related to textural variations in the samples and observe that changes related to the compaction of the samples are not relevant for the corresponding permeability predictions. We find that the proposed model does indeed provide an adequate prediction of the overall trend of the observed permeability values, but underestimates their actual values by approximately one order-of-magnitude. This discrepancy in turn points to the potential importance of phenomena, which are currently not accounted for in the model and which tend to reduce the characteristic size of the prevailing polarization cells compared to the considered model, such as, for example, membrane polarization, contacts of double-layers of neighbouring grains, and incorrect estimation of the size of the polarized cells because of the irregularity of natural sand grains.

The correlation between inbreeding and fitness: does allele size matter?

The availability of highly polymorphic genetic markers, in particular microsatellites, has made it possible to test the effect of inbreeding on fitness in the field and in the absence of pedigree information. It has been suggested that the squared difference in allele size at a locus (d(2)) might be a better indicator of the level of inbreeding than is heterozygosity. Using an elegant new analytical model, Tsitrone et al. now put this idea to the test, and to rest.

On dome-shaped norms of reaction for size-to-age at maturity in fishes

The optimal size-to-age at maturity depends on growth and mortality rates, which vary with environment. Therefore, organisms in spatially or temporaly changing environments should develop adaptative phenotypic plasticity for this trait. Experimental work by Alm (1959) on several fish species shows a dome-shape norm of reaction for size-to-age at maturity: size at maturity is smaller in both fast-growing and slow-growing fishes, than it is in fish with a medium growth rate. Using computer simulations, we show that such a dome-shaped norm of reaction is optimal when assuming a finite life span and a negative relationship between production and survival rates. This latter assumption is supported by empirical data, as well as by physiological and emographic arguments.

Cybercafés de Bamako : les usages de l'Internet au prisme de la parenté et des classes d'âge

Au Mali, les technologies de l'information et de la communication ont suscité un vif engouement. Portés par l'ancien Président Alpha Oumar Konaré, de nombreux projets gouvernementaux ont vu le jour. Parallèlement, les cybercafés se sont multipliés : moins de dix en 1997, ils étaient presque plus de 120 en 2006. Si les possibilités de se connecter à domicile et sur son lieu de travail se sont développées, les cybercafés restent des espaces idéaux pour découvrir les pratiques numériques. Une façon d'étudier les usages de l'Internet est de s'intéresser aux liens qui se nouent autour de l'ordinateur. La situation est complexe à Bamako, car les personnes qui accompagnent les clients des cybercafés sont nombreuses. L'analyse de cette nébuleuse relationnelle est l'occasion de procéder à un double détour. Géographique tout d'abord, dans la mesure où l'anthropologue est conduit à quitter le cybercafé pour se rendre dans l'espace social plus vaste appelé Cyber. Disciplinaire ensuite puisque son regard se déplace progressivement du média vers l'étude des règles qui régissent la parenté et les classes d'âge.