Nearest Neighbor Classification Rule for Mixtures of Discrete and Continuous Random Variables

In this paper very simple nonparametric classification rule for mixtures of discrete and continuous random variables is described. It is based on the method of nearest neighbor proposed by Cover and Hart (1967). The bounds on the limit of the nearest neighbor rule risks are given. Both lower and upper bound depend on the Bayes risk and the loss function. Finally the method is compared with other existing methods on some practical data set.     

An Efficient Regression Type Estimator in Survey Sampling

For the estimation of population mean in simple random sampling, an efficient regression-type estimator is proposed which is more efficient than the conventional regression estimator and hence than mean per unit estimator, ratio and product estimators and many other estimators proposed by various authors. Some numerical examples are included for illustration.     

Distinguishing Drift and Selection Empirically: “The Great Snail Debate” of the 1950s

Biologists and philosophers have been extremely pessimistic about the possibility of demonstrating random drift in nature, particularly when it comes to distinguishing random drift from natural selection. However, examination of a historical case – Maxime Lamotte’s study of natural populations of the land snail, Cepaea nemoralis in the 1950s – shows that while some pessimism is warranted, it has been overstated. Indeed, by describing a unique signature for drift and showing that this signature obtained in the populations under study, Lamotte was able to make a good case for a significant role for␣drift. It may be difficult to disentangle the causes of drift and selection acting in a population, but it is not (always) impossible.     

Joint multiple quantitative trait loci mapping for allometries of body compositions and metabolic traits to body weights in broiler

In order to map quantitative trait loci (QTLs) for allometries of body compositions and metabolic traits in chicken, we phenotypically characterize the allometric growths of multiple body components and metabolic traits relative to BWs using joint allometric scaling models and then establish random regression models (RRMs) to fit genetic effects of markers and minor polygenes derived from the pedigree on the allometric scalings. Prior to statistically inferring the QTLs for the allometric scalings by solving the RRMs, the LASSO technique is adopted to rapidly shrink most of marker genetic effects to zero. Computer simulation analysis confirms the reliability and adaptability of the so-called LASSO-RRM mapping method. In the F₂ population constructed by multiple families, we formulate two joint allometric scaling models of body compositions and metabolic traits, in which six of nine body compositions are tested as significant, while six of eight metabolic traits are as significant. For body compositions, a total of 14 QTLs, of which 9 dominant, were detected to be associated with the allometric scalings of drumstick, fat, heart, shank, liver and spleen to BWs; while for metabolic traits, a total of 19 QTLs also including 9 dominant be responsible for the allometries of T4, IGFI, IGFII, GLC, INS, IGR to BWs. The detectable QTLs or highly linked markers can be used to regulate relative growths of the body components and metabolic traits to BWs in marker-assisted breeding of chickens.     

Effect of feeding boxes on the behavior of stereotyping amur tigers (Panthera tigris altaica) in the Zurich Zoo,Zurich, Switzerland

The stereotyped pacing shown by the two Amur tigers in the Zurich Zoo was hypothesized as being caused by permanently frustrated appetitive foraging behavior. Several electrically controlled feeding boxes were installed and access to each box was possible only twice a day for 15 min at semi‐random times. The boxes had to be opened actively by the tigers. Two trials were carried out: one with solitary confinement, and one with paired confinement. During box feeding, the female's stereotyped pacing was significantly reduced from 16% (solitary confinement, conventional feeding) and 7% (paired confinement, conventional feeding) to 1% (solitary confinement) and less than 0.01% (paired confinement) of the daily observed time. The female's sleeping increased significantly in both solitary and paired confinement. The male only showed a significant reduction in stereotyped pacing behavior when kept with the female (conventional feeding: 10%; box feeding: <0.01% of the daily observed time). On days with a box‐feeding regime in paired confinement, the male spent 25% (83 min) of the observed time with active behavior at the feeding boxes. The results support the hypothesis that permanently frustrated appetitive foraging behavior causes stereotyped pacing in adult tigers. Zoo Biol 21:573–584, 2002. © 2002 Wiley‐Liss, Inc.     

A genetic and spatial Bayesian analysis of mastitis resistance

A nationwide health card recording system for dairy cattle was introduced in Norway in 1975 (the Norwegian Cattle Health Services). The data base holds information on mastitis occurrences on an individual cow basis. A reduction in mastitis frequency across the population is desired, and for this purpose risk factors are investigated. In this paper a Bayesian proportional hazards model is used for modelling the time to first veterinary treatment of clinical mastitis, including both genetic and environmental covariates. Sire effects were modelled as shared random components, and veterinary district was included as an environmental effect with prior spatial smoothing. A non-informative smoothing prior was assumed for the baseline hazard, and Markov chain Monte Carlo methods (MCMC) were used for inference. We propose a new measure of quality for sires, in terms of their posterior probability of being among the, say 10% best sires. The probability is an easily interpretable measure that can be directly used to rank sires. Estimating these complex probabilities is straightforward in an MCMC setting. The results indicate considerable differences between sires with regards to their daughters disease resistance. A regional effect was also discovered with the lowest risk of disease in the south-eastern parts of Norway.     

Comparison of 60-Hz electric fields and incandescent light as aversive stimuli controlling the behavior of rats

Rats were exposed to two procedures which enabled them to press a lever to turn off a 90 or 100 kV/m 60-Hz electric field or, later in the study, illumination from an incandescent lamp. Under one procedure, a response turned off the stimulus for a fixed duration, after which the stimulus was turned on again. A response during the off-period restarted the fixed duration. None of the rats turned the field off reliably. Next, under an alternative procedure, pressing one lever turned the field off; pressing the other lever turned it back on; responding under those conditions differed little from that seen at 0 kV/m. Under both procedures, when illumination from an incandescent lamp served as the stimulus, each rat did turn the stimulus off, and performances varied with stimulus intensity. The results show that a 100 kV/m 60-Hz electric field is not sufficient to function as an aversive stimulus under two procedures where illumination from a lamp does function as an aversive stimulus.     

Motifier: An IgOme Profiler Based on Peptide Motifs Using Machine Learning

Antibodies provide a comprehensive record of the encounters with threats and insults to the immune system. The ability to examine the repertoire of antibodies in serum and discover those that best represent “discriminating features” characteristic of various clinical situations, is potentially very useful. Recently, phage display technologies combined with Next-Generation Sequencing (NGS) produced a powerful experimental methodology, coined “Deep-Panning”, in which the spectrum of serum antibodies is probed. In order to extract meaningful biological insights from the tens of millions of affinity-selected peptides generated by Deep-Panning, advanced bioinformatics algorithms are a must. In this study, we describe Motifier, a computational pipeline comprised of a set of algorithms that systematically generates discriminatory peptide motifs based on the affinity-selected peptides identified by Deep-Panning. These motifs are shown to effectively characterize antibody binding activities and through the implementation of machine-learning protocols are shown to accurately classify complex antibody mixtures representing various biological conditions.     

A random fatigue of mechanize titanium abutment studied with Markoff chain and stochastic finite element formulation

To measure fatigue in dental implants and in its components, it is necessary to use a probabilistic analysis since the randomness in the output depends on a number of parameters (such as fatigue properties of titanium and applied loads, unknown beforehand as they depend on mastication habits). The purpose is to apply a probabilistic approximation in order to predict fatigue life, taking into account the randomness of variables. More accuracy on the results has been obtained by taking into account different load blocks with different amplitudes, as happens with bite forces during the day and allowing us to know how effects have different type of bruxism on the piece analysed.