Kernel smoothing is one of the most widely used non-parametric data smoothing techniques. We introduce a new R package ks for multivariate kernel smoothing. Currently it contains functionality for kernel density estimation and kernel discriminant analysis. It is a comprehensive package for bandwidth matrix selection, implementing a wide range of data-driven diagonal and unconstrained bandwidth selectors.

The nonrandom positioning of genes inside eukaryotic cell nuclei is implicated in central nuclear functions. However, the spatial organization of the genome remains largely uncharted, owing to limited resolution of optical microscopy, paucity of nuclear landmarks and moderate cell sampling. We developed a computational imaging approach that creates high-resolution probabilistic maps of subnuclear domains occupied by individual loci in budding yeast through automated analysis of thousands of living cells. After validation, we applied the technique to genes involved in galactose metabolism and ribosome biogenesis. We found that genomic loci are confined to 'gene territories' much smaller than the nucleus, which can be remodeled during transcriptional activation, and that the nucleolus is an important landmark for gene positioning. The technique can be used to visualize and quantify territory positions relative to each other and to nuclear landmarks, and should advance studies of nuclear architecture and function.

Anesthetics, widely used in magnetic resonance imaging (MRI) studies to avoid movement artifacts, could have profound effects on cerebral blood flow (CBF) and cerebrovascular coupling relative to the awake condition. Quantitative CBF and tissue oxygenation (blood oxygen level-dependent [BOLD]) were measured, using the continuous arterial-spin-labeling technique with echo-planar-imaging acquisition, in awake and anesthetized (2% isoflurane) rats under basal and hypercapnic conditions. All basal blood gases were within physiologic ranges. Blood pressure, respiration, and heart rates were within physiologic ranges in the awake condition but were depressed under anesthesia (P < 0.05). Regional CBF was heterogeneous with whole-brain CBF values of 0.86 +/- 0.25 and 1.27 +/- 0.29 mL. g-1. min-1 under awake and anesthetized conditions, respectively. Surprisingly, CBF was markedly higher (20% to 70% across different brain conditions) under isoflurane-anesthetized condition compared with the awake state (P < 0.01). Hypercapnia decreased pH, and increased Pco(2) and Po(2). During 5% CO(2) challenge, under awake and anesthetized conditions, respectively, CBF increased 51 +/- 11% and 25 +/- 4%, and BOLD increased 7.3 +/- 0.7% and 5.4 +/- 0.4%. During 10% CO(2) challenge, CBF increased 158 +/- 28% and 47 +/- 11%, and BOLD increased 12.5 +/- 0.9% and 7.2 +/- 0.5%. Since CBF and BOLD responses were substantially higher under awake condition whereas blood gases were not statistically different, it was concluded that cerebrovascular reactivity was suppressed by anesthetics. This study also shows that perfusion and perfusion-based functional MRI can be performed in awake animals.

Physical interactions between distinct chromosomal genomic loci are important for genomic functions including recombination and gene expression, but the mechanisms by which these interactions occur remain obscure. Using telomeric association as a model system, we analyzed here the in vivo organization of chromosome ends of haploid yeast cells during interphase. We separately labeled most of the 32 subtelomeres and analyzed their positions both in nuclear space and relative to three representative reference subtelomeres by high-throughput 3D microscopy and image processing. We show that subtelomeres are positioned nonrandomly at the nuclear periphery, depending on the genomic size of their chromosome arm, centromere attachment to the microtubule organizing center (spindle pole body, SPB), and the volume of the nucleolus. The distance of subtelomeres to the SPB increases consistently with chromosome arm length up to ≈300 kb; for larger arms the influence of chromosome arm length is weaker, but the effect of the nucleolar volume is stronger. Distances between pairs of subtelomeres also exhibit arm-length dependence and suggest, together with dynamic tracking experiments, that potential associations between subtelomeres are unexpectedly infrequent and transient. Our results suggest that interactions between subtelomeres are nonspecific and instead governed by physical constraints, including chromosome structure, attachment to the SPB, and nuclear crowding.genome organization | telomeric foci | image analysis

The performance of multivariate kernel density estimates depends crucially on the choice of bandwidth matrix, but progress towards developing good bandwidth matrix selectors has been relatively slow. In particular, previous studies of cross-validation (CV) methods have been restricted to biased and unbiased CV selection of diagonal bandwidth matrices. However, for certain types of target density the use of full (i.e. unconstrained) bandwidth matrices offers the potential for significantly improved density estimation. In this paper, we generalize earlier work from diagonal to full bandwidth matrices, and develop a smooth cross-validation (SCV) methodology for multivariate data. We consider optimization of the SCV technique with respect to a pilot bandwidth matrix. All the CV methods are studied using asymptotic analysis, simulation experiments and real data analysis. The results suggest that SCV for full bandwidth matrices is the most reliable of the CV methods. We also observe that experience from the univariate setting can sometimes be a misleading guide for understanding bandwidth selection in the multivariate case. Copyright 2005 Board of the Foundation of the Scandinavian Journal of Statistics..