Estimation of sequencing error rates in short reads.

BACKGROUND: Short-read data from next-generation sequencing technologies are now being generated across a range of research projects. The fidelity of this data can be affected by several factors and it is important to have simple and reliable approaches for monitoring it at the level of individual experiments. RESULTS: We developed a fast, scalable and accurate approach to estimating error rates in short reads, which has the added advantage of not requiring a reference genome. We build on the fundamental observation that there is a linear relationship between the copy number for a given read and the number of erroneous reads that differ from the read of interest by one or two bases. The slope of this relationship can be transformed to give an estimate of the error rate, both by read and by position. We present simulation studies as well as analyses of real data sets illustrating the precision and accuracy of this method, and we show that it is more accurate than alternatives that count the difference between the sample of interest and a reference genome. We show how this methodology led to the detection of mutations in the genome of the PhiX strain used for calibration of Illumina data. The proposed method is implemented in an R package, which can be downloaded from http://bcb.dfci.harvard.edu/∼vwang/shadowRegression.html. CONCLUSIONS: The proposed method can be used to monitor the quality of sequencing pipelines at the level of individual experiments without the use of reference genomes. Furthermore, having an estimate of the error rates gives one the opportunity to improve analyses and inferences in many applications of next-generation sequencing data.

A simulation-based approach for evaluating microarray analyses.

The biological complexity of gene expression makes simulation of gene expression data difficult. We propose a spike-in simulation that adds a single simulated gene to the data set of interest. Features of this spike-in gene may be manipulated to observe how often the spiked-in gene appears in the list of differentially expressed genes. This approach provides insight into the data analysis method, the observed data, and the manner in which the method and data interact without relying on indefensible assumptions regarding gene coexpression.

A review of orbital and intracranial magnetic resonance imaging in 79 canine and 13 feline patients (2004-2010).

OBJECTIVE: To review the distribution of orbital and intracranial disease in canine and feline patients undergoing magnetic resonance imaging (MRI) following referral to a veterinary ophthalmologist and to correlate results of MRI with pathologic conditions including neoplasia, suspected optic neuritis (ON) and orbital cellulitis. Recognized and emerging imaging techniques are reviewed. PROCEDURE: Medical records of 79 canine and 13 feline patients were reviewed. RESULTS: Neoplasia was diagnosed in 53/92 (57.6%) of patients. The most prevalent types of neoplasia were carcinoma (16/53, 30.1%), sarcoma (11/53, 20.8%), lymphoma (8/53, 15.1%) and presumptive meningioma (9/53, 17.0%). Carcinomas and sarcomas were characterized by bony lysis and intracranial/sinonasal extension. Lymphoma was generally unilateral, less invasive and originated from the ventromedial orbit. Intracranial masses representing presumptive meningiomas frequently exhibited a 'dural tail' sign. Diagnosis of suspected ON was made in 13 of 92 (14.1%) patients. Results of MRI in patients with suspected ON included unilateral optic nerve hyperintensity (3/13, 23.0%), bilateral optic nerve hyperintensity (1/13, 7.7%) and optic chiasmal hyperintensity (3/13, 23.0%). Seven suspected ON patients demonstrated intracranial multifocal patchy contrast enhancement (7/13, 53.8%). Diagnosis of orbital cellulitis was made in 12/92 (13.0%) patients. CONCLUSIONS: Orbital neoplasia was the most common pathologic condition detected. Essential Roentgen characteristics are helpful when diagnosing pathologic processes and providing prognoses in cases of orbital or intracranial disease. Magnetic resonance imaging comprises an important diagnostic component in cases of suspected ON. Emerging contrast and functional MRI techniques as well as SI data may increase our ability to characterize disease processes.

Long-Term Cognitive Decline After Newly Diagnosed Heart Failure: Longitudinal Analysis in the CHS (Cardiovascular Health Study).

BACKGROUND: Heart failure (HF) is associated with cognitive impairment. However, we know little about the time course of cognitive change after HF diagnosis, the importance of comorbid atrial fibrillation, or the role of ejection fraction. We sought to determine the associations of incident HF with rates of cognitive decline and whether these differed by atrial fibrillation status or reduced versus preserved ejection fraction. METHODS AND RESULTS: Participants were 4864 men and women aged ≥65 years without a history of HF and free of clinical stroke in the CHS (Cardiovascular Health Study)-a community-based prospective cohort study in the United States, with cognition assessed annually from 1989/1990 through 1998/1999. We identified 496 participants with incident HF by review of hospital discharge summaries and Medicare claims data, with adjudication according to standard criteria. Global cognitive ability was measured by the Modified Mini-Mental State Examination. In adjusted models, 5-year decline in model-predicted mean Modified Mini-Mental State Examination score was 10.2 points (95% confidence interval, 8.6-11.8) after incident HF diagnosed at 80 years of age, compared with a mean 5-year decline of 5.8 points (95% confidence interval, 5.3-6.2) from 80 to 85 years of age without HF. The association was stronger at older ages than at younger ages, did not vary significantly in the presence versus absence of atrial fibrillation (P=0.084), and did not vary significantly by reduced versus preserved ejection fraction (P=0.734). CONCLUSIONS: Decline in global cognitive ability tends to be faster after HF diagnosis than without HF. Clinical and public health implications of this finding warrant further attention.

Relationships and differentially expressed genes among pancreatic cancers examined by large-scale serial analysis of gene expression.

Pancreatic adenocarcinoma is among the most fatal of cancers, in part because of late diagnosis and a lack of effective therapies. Comprehensive studies are needed to better understand and address the cellular mechanisms and pathways of tumorigenesis. Serial analysis of gene expression was used to analyze gene expression profiles of pancreatic cancer cell lines, short-term cultures of normal pancreatic ductal epithelium, and primary pancreatic cancer tissue. A total of 294,920 tags representing 77,746 genes in 10 serial analysis of gene expression libraries were analyzed. A pancreatic cancer cell line (Hs766T) that exhibited a "normoid" profile of gene expression was identified. Several genes that may be involved in the fundamental nature of malignant changes in pancreatic ductal epithelium were suggested from those differentially and highly expressed in pancreatic cancer cells as compared with normal epithelium. Some overexpressed genes, such as S100A4, prostate stem cell antigen, carcinoembryonic antigen-related cell adhesion molecule 6, and mesothelin, suggest potential use as diagnostic markers. Others suggest potential novel therapeutic targets.

Proteotypic classification of spontaneous and transgenic mammary neoplasms.

INTRODUCTION: Mammary tumors in mice are categorized by using morphologic and architectural criteria. Immunolabeling for terminal differentiation markers was compared among a variety of mouse mammary neoplasms because expression of terminal differentiation markers, and especially of keratins, provides important information on the origin of neoplastic cells and their degree of differentiation. METHODS: Expression patterns for terminal differentiation markers were used to characterize tumor types and to study tumor progression in transgenic mouse models of mammary neoplasia (mice overexpressing Neu (Erbb2), Hras, Myc, Notch4, SV40-TAg, Tgfa, and Wnt1), in spontaneous mammary carcinomas, and in mammary neoplasms associated with infection by the mouse mammary tumor virus (MMTV). RESULTS: On the basis of the expression of terminal differentiation markers, three types of neoplasm were identified: first, simple carcinomas composed exclusively of cells with a luminal phenotype are characteristic of neoplasms arising in mice transgenic for Neu, Hras, Myc, Notch4, and SV40-TAg; second, 'complex carcinomas' displaying luminal and myoepithelial differentiation are characteristic of type P tumors arising in mice transgenic for Wnt1, neoplasms arising in mice infected by the MMTV, and spontaneous adenosquamous carcinomas; and third, 'carcinomas with epithelial to mesenchymal transition (EMT)' are a characteristic feature of tumor progression in Hras-, Myc-, and SV40-TAg-induced mammary neoplasms and PL/J and SJL/J mouse strains, and display de novo expression of myoepithelial and mesenchymal cell markers. In sharp contrast, EMT was not detected in papillary adenocarcinomas arising in BALB/cJ mice, spontaneous adenoacanthomas, neoplasms associated with MMTV-infection, or in neoplasms arising in mice transgenic for Neu and Wnt1. CONCLUSIONS: Immunohistochemical profiles of complex neoplasms are consistent with a stem cell origin, whereas simple carcinomas might originate from a cell committed to the luminal lineage. In addition, these results suggest that the initiating oncogenic events determine the morphologic features associated with cancer progression because EMT is observed only in certain types of neoplasm.

Improved statistical tests for differential gene expression by shrinking variance components estimates.

Combining information across genes in the statistical analysis of microarray data is desirable because of the relatively small number of data points obtained for each individual gene. Here we develop an estimator of the error variance that can borrow information across genes using the James-Stein shrinkage concept. A new test statistic (FS) is constructed using this estimator. The new statistic is compared with other statistics used to test for differential expression: the gene-specific F test (F1), the pooled-variance F statistic (F3), a hybrid statistic (F2) that uses the average of the individual and pooled variances, the regularized t-statistic, the posterior odds statistic B, and the SAM t-test. The FS-test shows best or nearly best power for detecting differentially expressed genes over a wide range of simulated data in which the variance components associated with individual genes are either homogeneous or heterogeneous. Thus FS provides a powerful and robust approach to test differential expression of genes that utilizes information not available in individual gene testing approaches and does not suffer from biases of the pooled variance approach.

A postgenomic method for predicting essential genes at subsaturation levels of mutagenesis: application to Mycobacterium tuberculosis.

We describe a postgenomic in silico approach for identifying genes that are likely to be essential and estimate their proportion in haploid genomes. With the knowledge of all sites eligible for mutagenesis and an experimentally determined partial list of nonessential genes from genome mutagenesis, a Bayesian statistical method provides reasonable predictions of essential genes with a subsaturation level of random mutagenesis. For mutagenesis, a transposon such as Himar1 is suitable as it inserts randomly into TA sites. All of the possible insertion sites may be determined a priori from the genome sequence and with this information, data on experimentally hit TA sites may be used to predict the proportion of genes that cannot be mutated. As a model, we used the Mycobacterium tuberculosis genome. Using the Himar1 transposon, we created a genetically defined collection of 1,425 insertion mutants. Based on our Bayesian statistical analysis using Markov chain Monte Carlo and the observed frequencies of transposon insertions in all of the genes, we estimated that the M. tuberculosis genome contains 35% (95% confidence interval, 28%-41%) essential genes. This analysis further revealed seven functional groups with high probabilities of being enriched in essential genes. The PE-PGRS (Pro-Glu polymorphic GC-rich repetitive sequence) family of genes, which are unique to mycobacteria, the polyketide/nonribosomal peptide synthase family, and mycolic and fatty acid biosynthesis gene families were disproportionately enriched in essential genes. At subsaturation levels of mutagenesis with a random transposon such as Himar1, this approach permits a statistical prediction of both the proportion and identities of essential genes of sequenced genomes.