Mixture models for analysis of melting temperature data

Background  

In addition to their use in detecting undesired real-time PCR products, melting temperatures are useful for detecting variations in the desired target sequences. Methodological improvements in recent years allow the generation of high-resolution melting-temperature (T_m) data. However, there is currently no convention on how to statistically analyze such high-resolution T_m data.     

Analysis of competing risks survival data when some failure types are missing

We propose a method to analyse competing risks survival datawhen failure types are missing for some individuals. Our approachis based on a standard proportional hazards structure for eachof the failure types, and involves the solution to estimatingequations. We present consistent and asymptotically normal estimatorsof the regression coefficients and related score tests. An appealingfeature is that individuals with known failure types make thesame contributions as they would to a standard proportionalhazards analysis. Contributions of individuals with unknownfailure types are weighted according to the probability thatthey failed from the cause of interest. Efficiency and robustnessare discussed. Results are illustrated with data from a breastcancer trial.     

Mixture models for assessing differential expression in complex tissues using microarray data

MOTIVATION: The use of DNA microarrays has become quite popular in many scientific and medical disciplines, such as in cancer research. One common goal of these studies is to determine which genes are differentially expressed between cancer and healthy tissue, or more generally, between two experimental conditions. A major complication in the molecular profiling of tumors using gene expression data is that the data represent a combination of tumor and normal cells. Much of the methodology developed for assessing differential expression with microarray data has assumed that tissue samples are homogeneous. RESULTS: In this paper, we outline a general framework for determining differential expression in the presence of mixed cell populations. We consider study designs in which paired tissues and unpaired tissues are available. A hierarchical mixture model is used for modeling the data; a combination of methods of moments procedures and the expectation-maximization algorithm are used to estimate the model parameters. The finite-sample properties of the methods are assessed in simulation studies; they are applied to two microarray datasets from cancer studies. Commands in the R language can be downloaded from the URL http://www.sph.umich.edu/~ghoshd/COMPBIO/COMPMIX/.     

Bayesian models for the analysis of genetic structure when populations are correlated

MOTIVATION: Population allele frequencies are correlated when populations have a shared history or when they exchange genes. Unfortunately, most models for allele frequency and inference about population structure ignore this correlation. Recent analytical results show that among populations, correlations can be very high, which could affect estimates of population genetic structure. In this study, we propose a mixture beta model to characterize the allele frequency distribution among populations. This formulation incorporates the correlation among populations as well as extending the model to data with different clusters of populations. RESULTS: Using simulated data, we show that in general, the mixture model provides a good approximation of the among-population allele frequency distribution and a good estimate of correlation among populations. Results from fitting the mixture model to a dataset of genotypes at 377 autosomal microsatellite loci from human populations indicate high correlation among populations, which may not be appropriate to neglect. Traditional measures of population structure tend to overestimate the amount of genetic differentiation when correlation is neglected. Inference is performed in a Bayesian framework. CONTACT: fur@ohsu.edu.     

Process of building biologically based dose-response models for developmental defects.

The problem of developing biologically-based dose-response models is addressed for predicting the prevalence of birth defects at low doses of toxic chemicals administered during pregnancy. To illustrate the process of incorporating biological information, a model is postulated to predict the prevalence of cleft palate for a chemical that reduces embryonic/fetal growth, which results in inadequate palatal cells for closure. Experimental bioassay data examining the prevalence of cleft palate in mice exposed to the herbicide 2,4,5-T are used to illustrate the process. With the limited data available, it is necessary to assume a model for cell growth and the relationship between the cell growth rate parameter and dose of 2,4,5-T. Also, a relationship between cleft palate prevalence and growth is assumed and then checked with experimental data. The purpose of the paper is not to provide a universal biologically based dose-response model for cleft palate, but rather to demonstrate the extent, and type of information and data required. It remains to be seen if the form of the model is appropriate for chemicals that primarily produce embryo/fetal malformations or death via reduced or delayed cellular growth.     

Estimation in capture-recapture models when covariates are subject to measurement errors

We consider estimation problems in capture-recapture models when the covariates or the auxiliary variables are measured with errors. The naive approach, which ignores measurement errors, is found to be unacceptable in the estimation of both regression parameters and population size: it yields estimators with biases increasing with the magnitude of errors, and flawed confidence intervals. To account for measurement errors, we derive a regression parameter estimator using a regression calibration method. We develop modified estimators of the population size accordingly. A simulation study shows that the resulting estimators are more satisfactory than those from either the naive approach or the simulation extrapolation (SIMEX) method. Data from a bird species Prinia flaviventris in Hong Kong are analyzed with and without the assumption of measurement errors, to demonstrate the effects of errors on estimations.     

Strength gains in obese females are unaffected by moderate dietary restriction

This study examined the effects of dietary restriction on strength gains from whole body resistance training. Comparisons were made between diet-restricted (n = 12) and non-diet-restricted (n = 10) obese women (mean +/- SD, 36.7 +/- 7.0% fat) undergoing identical 8-week resistance training regimens. Diet-restricted subjects reduced their dietary intake by 4200 kJ/day and reduced body mass by 3.9 kg over 8 weeks. Ten-repetition maximum masses were compared between the groups on biweekly intervals. Results indicated no differences between the groups with respect to the rate or magnitude of strength gains for any of the eight exercises. Significant pre- to post-test increases in strength (p less than 0.05) were found for all eight exercises. The rate or magnitude of strength gains induced by resistance training does not appear to be affected by moderate dietary restrictions in obese females.     

Vasoconstrictor mechanisms in the cerebral circulation are unaffected by insulin resistance

Insulin-resistance (IR) impairs agonist-induced relaxation in cerebral arteries, but little is known about its effect on constrictor mechanisms. We examined the vascular responses of the basilar artery (BA) and its side branches in anesthetized Zucker lean (ZL) and IR Zucker obese (ZO) rats using a cranial window technique. Endothelin-1 (ET-1) constricted the BAs in both the ZL and ZO rats, but there was no significant difference between the two groups (ZL: 36 +/- 8%; ZO: 33 +/- 3% at 10(-8) M). Inhibition of the ET(A) receptors by BQ-123 slightly increased the diameters of the BAs, with no difference shown between the ZL (6 +/- 1%) and ZO (5 +/- 3%) rats. Expressions of the ET(A) receptors and ET-1 mRNA examined by immunoblot analysis and RT-PCR, respectively, were also similar in the ZL and ZO groups. Phorbol 12,13-dibutyrate (PDBu), an activator of protein kinase C (PKC), and the thromboxane A(2) (TxA(2)) mimetic U-46619 constricted the BAs, but similarly to ET-1, there was no significant difference between the ZL and ZO groups (10(-6) M PDBu: ZL: 33 +/- 2%; ZO: 32 +/- 4%; and 10(-7) M U-46619: ZL: 23 +/- 1%; ZO: 19 +/- 2%). Inhibition of Rho-kinase with Y-27632 induced dilation of the BAs, and these responses were also comparable in the ZL and ZO rats (ZL: 39 +/- 4%; ZO: 38 +/- 2% at 10(-5) M). In contrast, nitric oxide-dependent relaxation to bradykinin was significantly reduced in the ZO rats (10(-6) M: 10 +/- 3%) compared with ZLs (29 +/- 7%, P < 0.01). These findings indicate that vasoconstrictor responses of the BA mediated by ET-1, TxA(2), PKC, and Rho-kinase are not affected by IR.     

Single-channel dose-response studies in single, cell-attached patches.

A method for carrying out dose-response studies of ion channel currents in cell-attached patches has been devised. Patch pipettes are filled at the tip with a solution containing one concentration of ligand and then backfilled with another. The concentration of ligand at the membrane is described as a function of time by the equation for diffusion in a cone, allowing response vs. time data to be transformed into a dose-response curve. For Xenopus myocyte cholinergic receptors, examples of the use of this method are given for several concentration-dependent reactions including blockade by the local anesthetic QX-222, activation by acetylcholine, and modulation of current amplitude by sodium ions. Several methods of analyzing the nonstationary channel kinetics are presented, including a pseudo-stationary approach that uses interval likelihood maximization.     

Flexible dose-response models for Japanese atomic bomb survivor data: Bayesian estimation and prediction of cancer risk

Generalised absolute risk models were fitted to the latest Japanese atomic bomb survivor cancer incidence data using Bayesian Markov Chain Monte Carlo methods, taking account of random errors in the DS86 dose estimates. The resulting uncertainty distributions in the relative risk model parameters were used to derive uncertainties in population cancer risks for a current UK population. Because of evidence for irregularities in the low-dose dose response, flexible dose-response models were used, consisting of a linear-quadratic-exponential model, used to model the high-dose part of the dose response, together with piecewise-linear adjustments for the two lowest dose groups. Following an assumed administered dose of 0.001 Sv, lifetime leukaemia radiation-induced incidence risks were estimated to be 1.11 x 10(-2) Sv(-1) (95% Bayesian CI -0.61, 2.38) using this model. Following an assumed administered dose of 0.001 Sv, lifetime solid cancer radiation-induced incidence risks were calculated to be 7.28 x 10(-2) Sv(-1) (95% Bayesian CI -10.63, 22.10) using this model. Overall, cancer incidence risks predicted by Bayesian Markov Chain Monte Carlo methods are similar to those derived by classical likelihood-based methods and which form the basis of established estimates of radiation-induced cancer risk.     

A marginalized conditional linear model for longitudinal binary data when informative dropout occurs in continuous time

Within the pattern-mixture modeling framework for informative dropout, conditional linear models (CLMs) are a useful approach to deal with dropout that can occur at any point in continuous time (not just at observation times). However, in contrast with selection models, inferences about marginal covariate effects in CLMs are not readily available if nonidentity links are used in the mean structures. In this article, we propose a CLM for long series of longitudinal binary data with marginal covariate effects directly specified. The association between the binary responses and the dropout time is taken into account by modeling the conditional mean of the binary response as well as the dependence between the binary responses given the dropout time. Specifically, parameters in both the conditional mean and dependence models are assumed to be linear or quadratic functions of the dropout time; and the continuous dropout time distribution is left completely unspecified. Inference is fully Bayesian. We illustrate the proposed model using data from a longitudinal study of depression in HIV-infected women, where the strategy of sensitivity analysis based on the extrapolation method is also demonstrated.