SURVIVAL PREDICTION FOR BRAIN TUMOR PATIENTS USING GENE EXPRESSION DATA

Brain tumor is one of the most aggressive types of cancer in humans, with an estimated median survival time of 12 months and only 4% of the patients surviving more than 5 years after disease diagnosis. Until recently, brain tumor prognosis has been based only on clinical information such as tumor grade and patient age, but there are reports indicating that molecular profiling of gliomas can reveal subgroups of patients with distinct survival rates. We hypothesize that coupling molecular profiling of brain tumors with clinical information might improve predictions of patient survival time and, consequently, better guide future treatment decisions. In order to evaluate this hypothesis, the general goal of this research is to build models for survival prediction of glioma patients using DNA molecular profiles (U133 Affymetrix gene expression microarrays) along with clinical information. First, a predictive Random Forest model is built for binary outcomes (i.e. short vs. long-term survival) and a small subset of genes whose expression values can be used to predict survival time is selected. Following, a new statistical methodology is developed for predicting time-to-death outcomes using Bayesian ensemble trees. Due to a large heterogeneity observed within prognostic classes obtained by the Random Forest model, prediction can be improved by relating time-to-death with gene expression profile directly. We propose a Bayesian ensemble model for survival prediction which is appropriate for high-dimensional data such as gene expression data. Our approach is based on the ensemble "sum-of-trees" model which is flexible to incorporate additive and interaction effects between genes. We specify a fully Bayesian hierarchical approach and illustrate our methodology for the CPH, Weibull, and AFT survival models. We overcome the lack of conjugacy using a latent variable formulation to model the covariate effects which decreases computation time for model fitting. Also, our proposed models provides a model-free way to select important predictive prognostic markers based on controlling false discovery rates. We compare the performance of our methods with baseline reference survival methods and apply our methodology to an unpublished data set of brain tumor survival times and gene expression data, selecting genes potentially related to the development of the disease under study. A closing discussion compares results obtained by Random Forest and Bayesian ensemble methods under the biological/clinical perspectives and highlights the statistical advantages and disadvantages of the new methodology in the context of DNA microarray data analysis.

Family business: Multiple members of major phytohormone classes orchestrate plant stress responses

Low-molecular-weight compounds such as jasmonic, abscisic and salicylic acids are commonly thought to be regulators of plant stress responses. However, it is becoming clear that these molecules, often referred to as phytohormones, are only a part of bigger groups of compounds with biological activity. We propose that the concept of "hormone families" may help to better understand plant physiological responses by taking into account not only the alleged main regulators, but also their precursors, conjugates and catabolites. Novel approaches to profile potentially active compounds in plants are discussed.

Up, down, strange and charm quark masses with Nf = 2+1+1 twisted mass lattice QCD

We present a lattice QCD calculation of the up, down, strange and charm quark masses performed using the gauge configurations produced by the European Twisted Mass Collaboration with Nf=2+1+1 dynamical quarks, which include in the sea, besides two light mass degenerate quarks, also the strange and charm quarks with masses close to their physical values. The simulations are based on a unitary setup for the two light quarks and on a mixed action approach for the strange and charm quarks. The analysis uses data at three values of the lattice spacing and pion masses in the range 210–450 MeV, allowing for accurate continuum limit and controlled chiral extrapolation. The quark mass renormalization is carried out non-perturbatively using the RI′-MOM method. The results for the quark masses converted to the scheme are: mud(2 GeV)=3.70(17) MeV, ms(2 GeV)=99.6(4.3) MeV and mc(mc)=1.348(46) GeV. We obtain also the quark mass ratios ms/mud=26.66(32) and mc/ms=11.62(16). By studying the mass splitting between the neutral and charged kaons and using available lattice results for the electromagnetic contributions, we evaluate mu/md=0.470(56), leading to mu=2.36(24) MeV and md=5.03(26) MeV.

A first look at maximally twisted mass lattice QCD calculations at the physical point

In this contribution, a first look at simulations using maximally twisted mass Wilson fermions at the physical point is presented. A lattice action including clover and twisted mass terms is presented and the Monte Carlo histories of one run with two mass-degenerate flavours at a single lattice spacing are shown. Measurements from the light and heavy-light pseudoscalar sectors are compared to previous Nf = 2 results and their phenomenological values. Finally, the strategy for extending simulations to Nf = 2+1+1 is outlined.

Classes of multichannel EEG microstates in light and deep hypnotic conditions

The study assessed the brain electric mechanisms of light and deep hypnotic conditions in the framework of EEG temporal microstates. Multichannel EEG of healthy volunteers during initial resting, light hypnosis, deep hypnosis, and eventual recovery was analyzed into temporal EEG microstates of four classes. Microstates are defined by the spatial configuration of their potential distribution maps ([Symbol: see text]potential landscapes') on the head surface. Because different potential landscapes must have been generated by different active neural assemblies, it is reasonable to assume that they also incorporate different brain functions. The observed four microstate classes were very similar to the four standard microstate classes A, B, C, D [Koenig, T. et al. Neuroimage, 2002;16: 41-8] and were labeled correspondingly. We expected a progression of microstate characteristics from initial resting to light to deep hypnosis. But, all three microstate parameters (duration, occurrence/second and %time coverage) yielded values for initial resting and final recovery that were between those of the two hypnotic conditions of light and deep hypnosis. Microstates of the classes B and D showed decreased duration, occurrence/second and %time coverage in deep hypnosis compared to light hypnosis; this was contrary to microstates of classes A and C which showed increased values of all three parameters. Reviewing the available information about microstates in other conditions, the changes from resting to light hypnosis in certain respects are reminiscent of changes to meditation states, and changes to deep hypnosis of those in schizophrenic states.