Multiple Lab Comparison of Microarray Platforms

Microarray technology is a powerful tool able to measure RNA expression for thousands of genes at once. Various studies have been published comparing competing platforms with mixed results: some find agreement, others do not. As the number of researchers starting to use microarrays and the number of crossplatform meta-analysis studies rapidly increase, appropriate platform assessments become more important. Here we present results from a comparison study that offers important improvements over those previously described in the literature. In particular, we notice that none of the previously published papers consider differences between labs. For this paper, a consortium of ten labs from the Washington DC/Baltimore (USA) area was formed to compare three heavily used platforms using identical RNA samples: Appropriate statistical analysis demonstrates that relatively large differences exist between labs using the same platform, but that the results from the best performing labs agree rather well. Supplemental material is available from http://www.biostat.jhsph.edu/~ririzarr/techcomp/

Estimating Multiple Tumor Transition Rates Based on Data from Survival/Sacrifice Experiments

In this paper, we focus on the model for two types of tumors. Tumor development can be described by four types of death rates and four tumor transition rates. We present a general semi-parametric model to estimate the tumor transition rates based on data from survival/sacrifice experiments. In the model, we make a proportional assumption of tumor transition rates on a common parametric function but no assumption of the death rates from any states. We derived the likelihood function of the data observed in such an experiment, and an EM algorithm that simplified estimating procedures. This article extends work on semi-parametric models for one type of tumor (see Portier and Dinse and Dinse) to two types of tumors.

Estimation of Transition Probabilities in Animal Carcinogenicity Experiments

This paper discusses estimation of the tumor incidence rate, the death rate given tumor is present and the death rate given tumor is absent using a discrete multistage model. The model was originally proposed by Dewanji and Kalbfleisch (1986) and the maximum likelihood estimate of the tumor incidence rate was obtained using EM algorithm. In this paper, we use a reparametrization to simplify the estimation procedure. The resulting estimates are not always the same as the maximum likelihood estimates but are asymptotically equivalent. In addition, an explicit expression for asymptotic variance and bias of the proposed estimators is also derived. These results can be used to compare efficiency of different sacrifice schemes in carcinogenicity experiments.

Development of a genotyping microarray for Usher syndrome

BACKGROUND: Usher syndrome, a combination of retinitis pigmentosa (RP) and sensorineural hearing loss with or without vestibular dysfunction, displays a high degree of clinical and genetic heterogeneity. Three clinical subtypes can be distinguished, based on the age of onset and severity of the hearing impairment, and the presence or absence of vestibular abnormalities. Thus far, eight genes have been implicated in the syndrome, together comprising 347 protein-coding exons. METHODS: To improve DNA diagnostics for patients with Usher syndrome, we developed a genotyping microarray based on the arrayed primer extension (APEX) method. Allele-specific oligonucleotides corresponding to all 298 Usher syndrome-associated sequence variants known to date, 76 of which are novel, were arrayed. RESULTS: Approximately half of these variants were validated using original patient DNAs, which yielded an accuracy of >98%. The efficiency of the Usher genotyping microarray was tested using DNAs from 370 unrelated European and American patients with Usher syndrome. Sequence variants were identified in 64/140 (46%) patients with Usher syndrome type I, 45/189 (24%) patients with Usher syndrome type II, 6/21 (29%) patients with Usher syndrome type III and 6/20 (30%) patients with atypical Usher syndrome. The chip also identified two novel sequence variants, c.400C>T (p.R134X) in PCDH15 and c.1606T>C (p.C536S) in USH2A. CONCLUSION: The Usher genotyping microarray is a versatile and affordable screening tool for Usher syndrome. Its efficiency will improve with the addition of novel sequence variants with minimal extra costs, making it a very useful first-pass screening tool.

Error models for microarray intensities

We derive the additive-multiplicative error model for microarray intensities, and describe two applications. For the detection of differentially expressed genes, we obtain a statistic whose variance is approximately independent of the mean intensity. For the post hoc calibration (normalization) of data with respect to experimental factors, we describe a method for parameter estimation.

Survival Analysis with Large Dimensional Covariates: An Application in Microarray Studies

Use of microarray technology often leads to high-dimensional and low- sample size data settings. Over the past several years, a variety of novel approaches have been proposed for variable selection in this context. However, only a small number of these have been adapted for time-to-event data where censoring is present. Among standard variable selection methods shown both to have good predictive accuracy and to be computationally efficient is the elastic net penalization approach. In this paper, adaptation of the elastic net approach is presented for variable selection both under the Cox proportional hazards model and under an accelerated failure time (AFT) model. Assessment of the two methods is conducted through simulation studies and through analysis of microarray data obtained from a set of patients with diffuse large B-cell lymphoma where time to survival is of interest. The approaches are shown to match or exceed the predictive performance of a Cox-based and an AFT-based variable selection method. The methods are moreover shown to be much more computationally efficient than their respective Cox- and AFT- based counterparts.

Fast DNA serotyping of Escherichia coli by use of an oligonucleotide microarray

Classical antibody-based serotyping of Escherichia coli is an important method in diagnostic microbiology for epidemiological purposes, as well as for a rough virulence assessment. However, serotyping is so tedious that its use is restricted to a few reference laboratories. To improve this situation we developed and validated a genetic approach for serotyping based on the microarray technology. The genes encoding the O-antigen flippase (wzx) and the O-antigen polymerase (wzy) were selected as target sequences for the O antigen, whereas fliC and related genes, which code for the flagellar monomer, were chosen as representatives for the H phenotype. Starting with a detailed bioinformatic analysis and oligonucleotide design, an ArrayTube-based assay was established: a fast and robust DNA extraction method was coupled with a site-specific, linear multiplex labeling procedure and hybridization analysis of the biotinylated amplicons. The microarray contained oligonucleotide DNA probes, each in duplicate, representing 24 of the epidemiologically most relevant of the over 180 known O antigens (O antigens 4, 6 to 9, 15, 26, 52, 53, 55, 79, 86, 91, 101, 103, 104, 111, 113, 114, 121, 128, 145, 157, and 172) as well as 47 of the 53 different H antigens (H antigens 1 to 12, 14 to 16, 18 to 21, 23 to 34, 37 to 43, 45, 46, 48, 49, 51 to 54, and 56). Evaluation of the microarray with a set of defined strains representing all O and H serotypes covered revealed that it has a high sensitivity and a high specificity. All of the conventionally typed 24 O groups and all of the 47 H serotypes were correctly identified. Moreover, strains which were nonmotile or nontypeable by previous serotyping assays yielded unequivocal results with the novel ArrayTube assay, which proved to be a valuable alternative to classical serotyping, allowing processing of single colonies within a single working day.

Microarray based study on virulence-associated genes and resistance determinants of Staphylococcus aureus isolates from cattle

Staphylococcus aureus is a common pathogen which can colonise and infect not only man, but also domestic animals. Especially, infection of cattle is of high economic relevance as S. aureus is an important causal agent of bovine mastitis. In the present contribution, a DNA microarray was applied for the study of 144 different gene targets, including resistance genes and genes encoding exotoxins, in S. aureus isolated from cows. One hundred and twenty-eight isolates from Germany and Switzerland were tested. These isolates were assigned to 20 different strains and nine clonal complexes. The majority of isolates belonged either to apparently closely related clonal complexes 8, 25, and 97 (together 34.4%) or were related to the sequenced bovine strain RF122 (48.4%). Notable characteristics of S. aureus of bovine origin are the carriage of intact haemolysin beta (in 82% of isolates tested), the absence of staphylokinase (in 89.1%), the presence of allelic variants of several exotoxins such as toxic shock syndrome toxin and enterotoxin N, and the occurrence of the leukocidin lukF-P83/lukM (in 53.1%). Two isolates were methicillin-resistant S. aureus (MRSA). One of them was a clonal complex 8 MRSA related to the epidemic MRSA strain Irish 01. The other one belonged to ST398/spa-type 34 resembling a newly emerging MRSA strain which has been described to occur in humans as well as in domestic animals. The presence of these two strains highlights the possibility of transfers of S. aureus strains between different host species.

Identification of markers to characterize and sort human articular chondrocytes with enhanced in vitro chondrogenic capacity

OBJECTIVE: To identify markers associated with the chondrogenic capacity of expanded human articular chondrocytes and to use these markers for sorting of more highly chondrogenic subpopulations. METHODS: The chondrogenic capacity of chondrocyte populations derived from different donors (n = 21) or different clonal strains from the same cartilage biopsy specimen (n = 21) was defined based on the glycosaminoglycan (GAG) content of tissues generated using a pellet culture model. Selected cell populations were analyzed by microarray and flow cytometry. In some experiments, cells were sorted using antibodies against molecules found to be associated with differential chondrogenic capacity and again assessed in pellet cultures. RESULTS: Significance Analysis of Microarrays indicated that chondrocytes with low chondrogenic capacity expressed higher levels of insulin-like growth factor 1 and of catabolic genes (e.g., matrix metalloproteinase 2, aggrecanase 2), while chondrocytes with high chondrogenic capacity expressed higher levels of genes involved in cell-cell or cell-matrix interactions (e.g., CD49c, CD49f). Flow cytometry analysis showed that CD44, CD151, and CD49c were expressed at significantly higher levels in chondrocytes with higher chondrogenic capacity. Flow cytometry analysis of clonal chondrocyte strains indicated that CD44 and CD151 could also identify more chondrogenic clones. Chondrocytes sorted for brighter CD49c or CD44 signal expression produced tissues with higher levels of GAG per DNA (up to 1.4-fold) and type II collagen messenger RNA (up to 3.4-fold) than did unsorted cells. CONCLUSION: We identified markers that allow characterization of the capacity of monolayer-expanded chondrocytes to form in vitro cartilaginous tissue and enable enrichment for subpopulations with higher chondrogenic capacity. These markers might be used as a means to predict and possibly improve the outcome of cell-based cartilage repair techniques.

Generic detection of coronaviruses and differentiation at the prototype strain level by reverse transcription-PCR and nonfluorescent low-density microarray

A nonfluorescent low-cost, low-density oligonucleotide array was designed for detecting the whole coronavirus genus after reverse transcription (RT)-PCR. The limit of detection was 15.7 copies/reaction. The clinical detection limit in patients with severe acute respiratory syndrome was 100 copies/sample. In 39 children suffering from coronavirus 229E, NL63, OC43, or HKU1, the sensitivity was equal to that of individual real-time RT-PCRs.