The morphological and physiological basis of delayed pollination overcoming pre-fertilization cross-incompatibility in Nicotiana.

Delayed pollination is widely used to overcome pre-fertilization incompatibility, but its regulatory mechanisms are unclear. When Nicotiana tabacum was cross-pollinated with pollen of N. alata, the incompatibility occurring in the basal 1/4 region of the style (pollinated at anthesis: 0-day-old pistil) was overcome by delayed pollination (of 6-day-old pistil), and the morphological changes and corresponding physiological basis are explored here. The structure and ultrastructure of the pistil were observed under fluorescence microscopy and transmission electron microscopy. Differentially expressed proteins were screened with a monoclonal antibody chip for Nicotiana, and protein expression and distribution were analysed by immunofluorescence. Cellulase and pectinase activities were tested using enzyme-linked immunosorbent assay kits. The style of Nicotiana is solid in the basal region and pollen tubes grow in the extracellular spaces (ECM) of the transmitting tissue (TTS) cells. Seven of the 22 identified proteins were cell wall-associated proteins and were expressed at higher levels during pistil senescence. Cellulase and pectinase activities increased. The TTS cells in the basal 1/4 region of the 0-day-old style were polygonal and tightly arranged, with narrow ECM, but these were oval or partially dissolved in the 6-day-old pistil, leading to wider ECM and richer secretions. The increased expression of cell wall proteins and enhanced enzyme activity during pistil senescence might partially be responsible for the cells becoming oval and the ECM enlarged, providing the morphological basis for delayed pollination overcoming the pre-fertilization incompatibility between N. tabacum and N. alata.

Prior robust empirical Bayes inference for large-scale data by conditioning on rank with application to microarray data.

Empirical Bayes methods have been extensively used for microarray data analysis by modeling the large number of unknown parameters as random effects. Empirical Bayes allows borrowing information across genes and can automatically adjust for multiple testing and selection bias. However, the standard empirical Bayes model can perform poorly if the assumed working prior deviates from the true prior. This paper proposes a new rank-conditioned inference in which the shrinkage and confidence intervals are based on the distribution of the error conditioned on rank of the data. Our approach is in contrast to a Bayesian posterior, which conditions on the data themselves. The new method is almost as efficient as standard Bayesian methods when the working prior is close to the true prior, and it is much more robust when the working prior is not close. In addition, it allows a more accurate (but also more complex) non-parametric estimate of the prior to be easily incorporated, resulting in improved inference. The new method's prior robustness is demonstrated via simulation experiments. Application to a breast cancer gene expression microarray dataset is presented. Our R package rank.Shrinkage provides a ready-to-use implementation of the proposed methodology.

Logistic regression for disease classification using microarray data: model selection in a large p and small n case.

MOTIVATION: Logistic regression is a standard method for building prediction models for a binary outcome and has been extended for disease classification with microarray data by many authors. A feature (gene) selection step, however, must be added to penalized logistic modeling due to a large number of genes and a small number of subjects. Model selection for this two-step approach requires new statistical tools because prediction error estimation ignoring the feature selection step can be severely downward biased. Generic methods such as cross-validation and non-parametric bootstrap can be very ineffective due to the big variability in the prediction error estimate. RESULTS: We propose a parametric bootstrap model for more accurate estimation of the prediction error that is tailored to the microarray data by borrowing from the extensive research in identifying differentially expressed genes, especially the local false discovery rate. The proposed method provides guidance on the two critical issues in model selection: the number of genes to include in the model and the optimal shrinkage for the penalized logistic regression. We show that selecting more than 20 genes usually helps little in further reducing the prediction error. Application to Golub's leukemia data and our own cervical cancer data leads to highly accurate prediction models. AVAILABILITY: R library GeneLogit at http://geocities.com/jg_liao

The effect of surgeon volume and hospital characteristics on in-hospital outcome after ureteral reimplantation in children.

The aim of this study was to examine the association between surgeon and hospital characteristics on in-hospital outcome after ureteral reimplantation in children. Patients<18 years undergoing vesicoureteral reimplantation (n=3,109) were identified in Kids' Inpatient Database, an administrative database containing discharge records from 27 states during 2000 in the US. Based on patient volume in 2000, surgeons were designated as low volume (<11 procedures), medium volume (11-20 procedures) and high volume (>20 procedures) surgeons. Length of stay and hospital charges were analyzed using multivariate linear regression analysis. A significant association between shorter length of stay and higher surgeon volume (p=0.02) was observed that was independent of children's hospital status, hospital volume and other hospital characteristics. Length of stay was 20% shorter when the procedure was performed by the highest volume surgeons compared to when performed by the lowest. No significant effect of surgeon volume on hospital charges, however, was observed. Higher surgeon volume was associated with shorter length of stay but no difference in hospital charges among children undergoing vesicoureteral reimplantation.

Effect of surgeon and hospital characteristics on outcome after pyloromyotomy.

BACKGROUND: Previous studies have suggested that the outcome after pyloromyotomy is improved with increased surgeon experience. Others have proposed that infants with pyloric stenosis are best treated by specialty-trained pediatric surgeons or at children's hospitals. HYPOTHESIS: Surgeon and hospital characteristics affect complications, length of stay, and hospital charges after pyloromyotomy. DESIGN: Data for a nationally representative sample of infants (n = 1277) who underwent pyloromyotomy in 2000 in the United States were obtained from the Kids' Inpatient Database. Surgeon and hospital volumes were stratified into quintiles. Multivariate analyses were performed to analyze the impact of surgeon and hospital volume on length of stay, charges, and major operative complications using models that accounted for the hierarchical structure of patient-, surgeon-, and hospital-level covariates. RESULTS: No association between surgeon volume and either length of stay or charges was observed. Higher surgeon volume, however, was associated with fewer complications (P<.001). Surgeons with the highest volume had a 90% lower risk of complications than those with the lowest volume. Higher hospital volume was associated with shorter length of stay (P<.001). No association between hospital volume and either charges or risk of complications was observed. CONCLUSIONS: Higher surgeon and hospital volumes are associated with better outcome among infants who are treated for pyloric stenosis. Identification of aspects of medical and surgical treatment that account for this finding may lead to improvement in the outcome of infants undergoing pyloromyotomy.

A mixture model for estimating the local false discovery rate in DNA microarray analysis.

MOTIVATION: Statistical methods based on controlling the false discovery rate (FDR) or positive false discovery rate (pFDR) are now well established in identifying differentially expressed genes in DNA microarray. Several authors have recently raised the important issue that FDR or pFDR may give misleading inference when specific genes are of interest because they average the genes under consideration with genes that show stronger evidence for differential expression. The paper proposes a flexible and robust mixture model for estimating the local FDR which quantifies how plausible each specific gene expresses differentially. RESULTS: We develop a special mixture model tailored to multiple testing by requiring the P-value distribution for the differentially expressed genes to be stochastically smaller than the P-value distribution for the non-differentially expressed genes. A smoothing mechanism is built in. The proposed model gives robust estimation of local FDR for any reasonable underlying P-value distributions. It also provides a single framework for estimating the proportion of differentially expressed genes, pFDR, negative predictive values, sensitivity and specificity. A cervical cancer study shows that the local FDR gives more specific and relevant quantification of the evidence for differential expression that can be substantially different from pFDR. AVAILABILITY: An R function implementing the proposed model is available at http://www.geocities.com/jg_liao/software

A hierarchical Bayesian model for combining multiple 2 x 2 tables using conditional likelihoods.

This paper introduces a hierarchical Bayesian model for combining multiple 2 x 2 tables that allows the flexibility of different odds ratio estimates for different tables and at the same time allows the tables to borrow information from each other. The proposed model, however, is different from a full Bayesian model in that the nuisance parameters are eliminated by conditioning instead of integration. The motivation is a more robust model and a faster and more stable Gibbs algorithm. We work out a Gibbs scheme using the adaptive rejection sampling for log concave density and an algorithm for the mean and variance of the noncentral hypergeometric distribution. The model is applied to a multicenter ulcer clinical trial.

Statistical modelling of the AIDS epidemic for forecasting health care needs.

The objective of this paper is to develop statistical methods for estimating current and future numbers of individuals in different stages of the natural history of the human immunodeficiency (AIDS) virus infection and to evaluate the impact of therapeutic advances on these numbers. The approach is to extend the method of back-calculation to allow for a multistage model of natural history and to permit the hazard functions of progression from one stage to the next to depend on calendar time. Quasi-likelihood estimates of key quantities for evaluating health care needs can be obtained through iteratively reweighted least squares under weakly parametric models for the infection rate. An approach is proposed for incorporating into the analysis independent estimates of human immunodeficiency virus (HIV) prevalence obtained from epidemiologic surveys. The methods are applied to the AIDS epidemic in the United States. Short-term projections are given of both AIDS incidence and the numbers of HIV-infected AIDS-free individuals with CD4 cell depletion. The impact of therapeutic advances on these numbers is evaluated using a change-point hazard model. A number of important sources of uncertainty must be considered when interpreting the results, including uncertainties in the specified hazard functions of disease progression, in the parametric model for the infection rate, in the AIDS incidence data, in the efficacy of treatment, and in the proportions of HIV-infected individuals receiving treatment.

The analysis of delays in disease reporting: methods and results for the acquired immunodeficiency syndrome.

In order to monitor accurately trends in disease incidence, it is necessary to account for delays in the reporting of cases to central registries. The objective of the paper is to develop simple methods for the analysis of reporting delays in order to identify the main sources of heterogeneity and to adjust reported disease incidence data. The analysis is complicated because the data are right truncated. A simple and flexible method for the regression analysis of reporting delays is proposed, which can be easily implemented with standard computing tools for generalized linear models or logistic regression. The method was used to analyze delays in reporting the acquired immunodeficiency syndrome in the United States among cases who met the pre-1987 surveillance definition. This analysis showed significant geographic variation. Delays were shortest in the Northeast and longest in the South. The influences of risk groups and calendar year of diagnosis were not consistent across each of the geographic regions. Variation among risk groups was attributed primarily to slower reporting of transfusion-associated and pediatric acquired immunodeficiency syndrome cases. An overall trend toward longer delays with calendar time of diagnosis was attributed primarily to a trend toward longer delays in the Northeast. These methods and results are useful both for the evaluation of surveillance procedures in order to improve disease reporting and for adjustment of disease incidence data.