Comparing cDNA and oligonucleotide array data: concordance of gene expression across platforms for the NCI-60 cancer cells

The non-coding RNA landscape of human hematopoiesis and leukemia

Non-coding RNAs have emerged as crucial regulators of gene expression and cell fate decisions. However, their expression patterns and regulatory functions during normal and malignant human hematopoiesis are incompletely understood. Here we present a comprehensive resource defining the non-coding RNA landscape of the human hematopoietic system. Based on highly specific non-coding RNA expression portraits per blood cell population, we identify unique fingerprint non-coding RNAs—such as LINC00173 in granulocytes—and assign these to critical regulatory circuits involved in blood homeostasis. Following the incorporation of acute myeloid leukemia samples into the landscape, we further uncover prognostically relevant non-coding RNA stem cell signatures shared between acute myeloid leukemia blasts and healthy hematopoietic stem cells. Our findings highlight the importance of the non-coding transcriptome in the formation and maintenance of the human blood hierarchy.

While micro-RNAs are known regulators of haematopoiesis and leukemogenesis, the role of long non-coding RNAs is less clear. Here the authors provide a non-coding RNA expression landscape of the human hematopoietic system, highlighting their role in the formation and maintenance of the human blood hierarchy.

Identification of differentially-expressed genes between early-stage adenocarcinoma and squamous cell carcinoma lung cancer using meta-analysis methods

Lung adenocarcinoma (AC) and squamous cell lung carcinoma (SCC) are two major subtypes of non-small cell lung cancer (NSCLC). Previous studies have demonstrated that fundamental differences exist in the underlying mechanisms of tumor development, growth and invasion between these subtypes. The investigation of differentially-expressed genes (DEGs) between these two NSCLC subtypes is useful for determining and understanding such differences. The present study aimed to identify those DEGs using meta-analysis and the data from four microarray experiments, consisting of 164 AC and 161 SCC samples. Raw gene expression values were converted into the probability of expression (POE) representing the differentially-expressed probability of a gene and expression barcode values representing its expression status. The results indicated that when applying a meta-analysis using barcode values, heterogeneity in genes across studies was less severe than when applying a meta-analysis using POE values. DEGs in each meta-analysis method overlapped substantially (P=1.3×10^-4), but the barcode method yielded a lower global false discovery rate. Based on this and several other performance statistics, it was concluded that the barcode approach outperformed the POE method. Finally, using those DEGs, ontology and pathway analyses were conducted. A number of genes and enriched pathways were found to be closely associated with NSCLC.

Development of a Drug-Response Modeling Framework to Identify Cell Line Derived Translational Biomarkers That Can Predict Treatment Outcome to Erlotinib or Sorafenib

Development of drug responsive biomarkers from pre-clinical data is a critical step in drug discovery, as it enables patient stratification in clinical trial design. Such translational biomarkers can be validated in early clinical trial phases and utilized as a patient inclusion parameter in later stage trials. Here we present a study on building accurate and selective drug sensitivity models for Erlotinib or Sorafenib from pre-clinical in vitro data, followed by validation of individual models on corresponding treatment arms from patient data generated in the BATTLE clinical trial. A Partial Least Squares Regression (PLSR) based modeling framework was designed and implemented, using a special splitting strategy and canonical pathways to capture robust information for model building. Erlotinib and Sorafenib predictive models could be used to identify a sub-group of patients that respond better to the corresponding treatment, and these models are specific to the corresponding drugs. The model derived signature genes reflect each drug’s known mechanism of action. Also, the models predict each drug’s potential cancer indications consistent with clinical trial results from a selection of globally normalized GEO expression datasets.

Retrospective analysis of survival improvement by molecular biomarker-based personalized chemotherapy for recurrent ovarian cancer

Aggressive tumors such as epithelial ovarian cancer (EOC) are highly heterogeneous in their therapeutic response, making it difficult to improve overall response by using drugs in unselected patients. The goal of this study was to retrospectively, but independently, examine whether biomarker-based personalized chemotherapy selection could improve survival of EOC patients. Using in vitro drug sensitivity and patient clinical outcome data, we have developed co-expression extrapolation (COXEN) biomarker models for predicting patient response to three standard chemotherapy drugs used to treat advanced EOC: paclitaxel, cyclophosphamide, and topotecan, for which sufficient patient data were available for our modeling and independent validation. Four different cohorts of 783 EOC patients were used in our study, including two cohorts of 499 patients for independent validation. The COXEN predictors for the three drugs independently showed high prediction both for patient short-term therapeutic response and long-term survival for recurrent EOC. We then examined the potential clinical benefit of the simultaneous use of the three drug predictors for a large diverse EOC cohort in a prospective manner, finding that the median overall survival was 21 months longer for recurrent EOC patients who were treated with the predicted most effective chemotherapies. Survival improvement was greater for platinum-sensitive patients if they were treated with the predicted most beneficial drugs. Following the FDA guidelines for diagnostic prediction analysis, our study has retrospectively, yet independently, showed a potential for biomarker-based personalized chemotherapy selection to significantly improve survival of patients in the heterogeneous EOC population when using standard chemotherapies.

Integrative biological analysis for neuropsychopharmacology

Although advances in psychotherapy have been made in recent years, drug discovery for brain diseases such as schizophrenia and mood disorders has stagnated. The need for new biomarkers and validated therapeutic targets in the field of neuropsychopharmacology is widely unmet. The brain is the most complex part of human anatomy from the standpoint of number and types of cells, their interconnections, and circuitry. To better meet patient needs, improved methods to approach brain studies by understanding functional networks that interact with the genome are being developed. The integrated biological approaches--proteomics, transcriptomics, metabolomics, and glycomics--have a strong record in several areas of biomedicine, including neurochemistry and neuro-oncology. Published applications of an integrated approach to projects of neurological, psychiatric, and pharmacological natures are still few but show promise to provide deep biological knowledge derived from cells, animal models, and clinical materials. Future studies that yield insights based on integrated analyses promise to deliver new therapeutic targets and biomarkers for personalized medicine.

Integrative correlation: Properties and relation to canonical correlations

The integrative correlation coefficient was developed to facilitate the validation of expression microarray results in public datasets, by identifying genes that are reproducibly measured across studies and even across microarray platforms. In the current study, we develop a number of interesting and important mathematical and statistical properties of the integrative correlation coefficient, including a unique permutation-based null distribution with the unusual property that the variance does not shrink as the sample size increases, discussing how these findings impact its use and interpretation, and what they have to say about any method for identifying reproducible genes in a meta-analysis.

Modulators of sensitivity and resistance to inhibition of PI3K identified in a pharmacogenomic screen of the NCI-60 human tumor cell line collection

The phosphoinositide 3-kinase (PI3K) signaling pathway is significantly altered in a wide variety of human cancers, driving cancer cell growth and survival. Consequently, a large number of PI3K inhibitors are now in clinical development. To begin to improve the selection of patients for treatment with PI3K inhibitors and to identify de novo determinants of patient response, we sought to identify and characterize candidate genomic and phosphoproteomic biomarkers predictive of response to the selective PI3K inhibitor, GDC-0941, using the NCI-60 human tumor cell line collection. In this study, sixty diverse tumor cell lines were exposed to GDC-0941 and classified by GI(50) value as sensitive or resistant. The most sensitive and resistant cell lines were analyzed for their baseline levels of gene expression and phosphorylation of key signaling nodes. Phosphorylation or activation status of both the PI3K-Akt signaling axis and PARP were correlated with in vitro response to GDC-0941. A gene expression signature associated with in vitro sensitivity to GDC-0941 was also identified. Furthermore, in vitro siRNA-mediated silencing of two genes in this signature, OGT and DDN, validated their role in modulating sensitivity to GDC-0941 in numerous cell lines and begins to provide biological insights into their role as chemosensitizers. These candidate biomarkers will offer useful tools to begin a more thorough understanding of determinants of patient response to PI3K inhibitors and merit exploration in human cancer patients treated with PI3K inhibitors.

Multi-gene expression predictors of single drug responses to adjuvant chemotherapy in ovarian carcinoma: predicting platinum resistance

Despite advances in radical surgery and chemotherapy delivery, ovarian cancer is the most lethal gynecologic malignancy. Standard therapy includes treatment with platinum-based combination chemotherapies yet there is no biomarker model to predict their responses to these agents. We here have developed and independently tested our multi-gene molecular predictors for forecasting patients' responses to individual drugs on a cohort of 55 ovarian cancer patients. To independently validate these molecular predictors, we performed microarray profiling on FFPE tumor samples of 55 ovarian cancer patients (UVA-55) treated with platinum-based adjuvant chemotherapy. Genome-wide chemosensitivity biomarkers were initially discovered from the in vitro drug activities and genomic expression data for carboplatin and paclitaxel, respectively. Multivariate predictors were trained with the cell line data and then evaluated with a historical patient cohort. For the UVA-55 cohort, the carboplatin, taxol, and combination predictors significantly stratified responder patients and non-responder patients (p = 0.019, 0.04, 0.014) with sensitivity = 91%, 96%, 93 and NPV = 57%, 67%, 67% in pathologic clinical response. The combination predictor also demonstrated a significant survival difference between predicted responders and non-responders with a median survival of 55.4 months vs. 32.1 months. Thus, COXEN single- and combination-drug predictors successfully stratified platinum resistance and taxane response in an independent cohort of ovarian cancer patients based on their FFPE tumor samples.

Assumption weighting for incorporating heterogeneity into meta-analysis of genomic data

MOTIVATION

There is now a large literature on statistical methods for the meta-analysis of genomic data from multiple studies. However, a crucial assumption for performing many of these analyses is that the data exhibit small between-study variation or that this heterogeneity can be sufficiently modelled probabilistically.

RESULTS

In this article, we propose 'assumption weighting', which exploits a weighted hypothesis testing framework proposed by Genovese et al. to incorporate tests of between-study variation into the meta-analysis context. This methodology is fast and computationally simple to implement. Several weighting schemes are considered and compared using simulation studies. In addition, we illustrate application of the proposed methodology using data from several high-profile stem cell gene expression datasets.

Comprehensive literature review and statistical considerations for microarray meta-analysis

With the rapid advances of various high-throughput technologies, generation of ‘-omics’ data is commonplace in almost every biomedical field. Effective data management and analytical approaches are essential to fully decipher the biological knowledge contained in the tremendous amount of experimental data. Meta-analysis, a set of statistical tools for combining multiple studies of a related hypothesis, has become popular in genomic research. Here, we perform a systematic search from PubMed and manual collection to obtain 620 genomic meta-analysis papers, of which 333 microarray meta-analysis papers are summarized as the basis of this paper and the other 249 GWAS meta-analysis papers are discussed in the next companion paper. The review in the present paper focuses on various biological purposes of microarray meta-analysis, databases and software and related statistical procedures. Statistical considerations of such an analysis are further scrutinized and illustrated by a case study. Finally, several open questions are listed and discussed.

TLX1 and NOTCH coregulate transcription in T cell acute lymphoblastic leukemia cells

BACKGROUND

The homeobox gene TLX1 (for T-cell leukemia homeobox 1, previously known as HOX11) is inappropriately expressed in a major subgroup of T cell acute lymphoblastic leukemia (T-ALL) where it is strongly associated with activating NOTCH1 mutations. Despite the recognition that these genetic lesions cooperate in leukemogenesis, there have been no mechanistic studies addressing how TLX1 and NOTCH1 functionally interact to promote the leukemic phenotype.

RESULTS

Global gene expression profiling after downregulation of TLX1 and inhibition of the NOTCH pathway in ALL-SIL cells revealed that TLX1 synergistically regulated more than 60% of the NOTCH-responsive genes. Structure-function analysis demonstrated that TLX1 binding to Groucho-related TLE corepressors was necessary for maximal transcriptional regulation of the NOTCH-responsive genes tested, implicating TLX1 modulation of the NOTCH-TLE regulatory network. Comparison of the dataset to publicly available biological databases indicated that the TLX1/NOTCH-coregulated genes are frequently targeted by MYC. Gain- and loss-of-function experiments confirmed that MYC was an essential mediator of TLX1/NOTCH transcriptional output and growth promotion in ALL-SIL cells, with TLX1 contributing to the NOTCH-MYC regulatory axis by posttranscriptional enhancement of MYC protein levels. Functional classification of the TLX1/NOTCH-coregulated targets also showed enrichment for genes associated with other human cancers as well as those involved in developmental processes. In particular, we found that TLX1, NOTCH and MYC coregulate CD1B and RAG1, characteristic markers of early cortical thymocytes, and that concerted downregulation of the TLX1 and NOTCH pathways resulted in their irreversible repression.

CONCLUSIONS

We found that TLX1 and NOTCH synergistically regulate transcription in T-ALL, at least in part via the sharing of a TLE corepressor and by augmenting expression of MYC. We conclude that the TLX1/NOTCH/MYC network is a central determinant promoting the growth and survival of TLX1+ T-ALL cells. In addition, the TLX1/NOTCH/MYC transcriptional network coregulates genes involved in T cell development, such as CD1 and RAG family members, and therefore may prescribe the early cortical stage of differentiation arrest characteristic of the TLX1 subgroup of T-ALL.

Comparison and consolidation of microarray data sets of human tissue expression

Background

Human tissue displays a remarkable diversity in structure and function. To understand how such diversity emerges from the same DNA, systematic measurements of gene expression across different tissues in the human body are essential. Several recent studies addressed this formidable task using microarray technologies. These large tissue expression data sets have provided us an important basis for biomedical research. However, it is well known that microarray data can be compromised by high noise level and various experimental artefacts. Critical comparison of different data sets can help to reveal such errors and to avoid pitfalls in their application.

Results

We present here the first comparison and integration of four freely available tissue expression data sets generated using three different microarray platforms and containing a total of 377 microarray hybridizations. When assessing the tissue expression of genes, we found that the results considerably depend on the chosen data set. Nevertheless, the comparison also revealed statistically significant similarity of gene expression profiles across different platforms. This enabled us to construct consolidated lists of platform-independent tissue-specific genes using a set of complementary measures. Follow-up analyses showed that results based on consolidated data tend to be more reliable.

Conclusions

Our study strongly indicates that the consolidation of the four different tissue expression data sets can increase data quality and can lead to biologically more meaningful results. The provided compendium of platform-independent gene lists should facilitate the identification of novel tissue-specific marker genes.

Analysis of expression of drug resistance-linked ABC transporters in cancer cells by quantitative RT-PCR

Quantitative real-time PCR (qRT-PCR) boasts many advantages over microarrays. For instance, very low amounts of total RNA are required to yield highly accurate and reproducible detection of transcript levels. As a consequence, qRT-PCR has become a popular technique for assessing gene expression levels and is now the gold standard. In this chapter, qRT-PCR using two distinct chemistries, SYBR Green and TaqMan, are described. We compare ABC transporter levels in various drug-resistant cancer cell lines by employing each method. SYBR Green yields reproducible results; nevertheless, TaqMan chemistry is superior to SYBR Green, as it displays higher specificity and sensitivity. Gene expression analysis by qRT-PCR is a powerful technique and shows potential as a diagnostic tool for predicting drug response in cancer patients.

Evaluation of cDNA microarray data by multiple clones mapping to the same transcript

Although novel technologies are rapidly emerging, the cDNA microarray data accumulated is still and will be an important source for bioinformatics and biological studies. Thus, the reliability and applicability of the cDNA microarray data warrants further evaluation. In cDNA microarrays, multiple clones are measured for a transcript, which can be exploited to evaluate the consistency of microarray data. We show that even for pairs of RCs, the average Pearson correlation coefficient of their measurements is not high. However, this low consistency could largely be explained by random noise signals for a fraction of unexpressed genes and/or low signal-to-noise ratios for low abundance transcripts. Encouragingly, a large fraction of inconsistent data will be filtered out in the procedure of selecting differentially expressed genes (DEGs). Therefore, although cDNA microarray data are of low consistency, applications based on DEGs selections could still reach correct biological results, especially at the functional modules level.

Correlations between RNA and protein expression profiles in 23 human cell lines

Background

The Central Dogma of biology holds, in famously simplified terms, that DNA makes RNA makes proteins, but there is considerable uncertainty regarding the general, genome-wide correlation between levels of RNA and corresponding proteins. Therefore, to assess degrees of this correlation we compared the RNA profiles (determined using both cDNA- and oligo-based microarrays) and protein profiles (determined immunohistochemically in tissue microarrays) of 1066 gene products in 23 human cell lines.

Results

A high mean correlation coefficient (0.52) was obtained from the pairwise comparison of RNA levels determined by the two platforms. Significant correlations, with correlation coefficients exceeding 0.445, between protein and RNA levels were also obtained for a third of the specific gene products. However, the correlation coefficients between levels of RNA and protein products of specific genes varied widely, and the mean correlations between the protein and corresponding RNA levels determined using the cDNA- and oligo-based microarrays were 0.25 and 0.20, respectively.

Conclusion

Significant correlations were found in one third of the examined RNA species and corresponding proteins. These results suggest that RNA profiling might provide indirect support to antibodies' specificity, since whenever a evident correlation between the RNA and protein profiles exists, this can sustain that the antibodies used in the immunoassay recognized their cognate antigens.

Identification of compounds that correlate with ABCG2 transporter function in the National Cancer Institute Anticancer Drug Screen

ABCG2 is an ATP-binding cassette transporter that counts multiple anticancer compounds among its substrates and is believed to regulate oral bioavailability as well as serve a protective role in the blood-brain barrier, the maternal-fetal barrier, and hematopoietic stem cells. We sought to determine whether novel compounds that interact with the transporter could be identified through analysis of cytotoxicity profiles recorded in the NCI Anticancer Drug Screen database. A flow cytometric assay was used to measure ABCG2 function in the 60 cell lines and generate a molecular profile for COMPARE analysis. This strategy identified >70 compounds with Pearson correlation coefficients (PCCs) >0.4, where reduced drug sensitivity correlated with ABCG2 expression, as well as >120 compounds with PCCs < -0.4, indicating compounds to which ABCG2 expression conferred greater sensitivity. Despite identification of known single nucleotide polymorphisms in the ABCG2 gene in a number of the cell lines, omission of these lines from the COMPARE analysis did not affect PCCs. Available compounds were subjected to validation studies to confirm interaction with the transporter, including flow cytometry, [(125)I]IAAP binding, and cytotoxicity assays, and interaction was documented in 20 of the 27 compounds studied. Although known substrates of ABCG2 such as mitoxantrone or topotecan were not identified, we characterized three novel substrates-5-hydroxypicolinaldehyde thiosemicarbazone (NSC107392), (E)-N-(1-decylsulfanyl-3-hydroxypropan-2-yl)-3-(6-methyl-2,4-dioxo-1H-pyrimidin-5-yl)prop-2-enamide (NSC265473), and 1,2,3,4,7-pentahydroxy-1,3,4,4a,5,11b-hexahydro[1,3]dioxolo[4,5-j]phenanthridin-6(2H)-one [NSC349156 (pancratistatin)]-and four compounds that inhibited transporter function-2-[methyl(2-pyridin-2-ylethyl)-amino]fluoren-9-one hydroiodide (NSC24048), 5-amino-6-(7-amino-5,8-dihydro-6-methoxy-5,8-dioxo-2-quinolinyl)-4-(2-hydroxy-3,4-dimethoxyphenyl)-3-methyl-2-pyridinecarboxylic acid, methyl ester (NSC45384), (17beta)-2,4-dibromo-estra-1,3,5(10)-triene-3,17-diol (NSC103054), and methyl N-(pyridine-4-carbonylamino)carbamodithioate (NSC636795). In summary, COMPARE analysis of the NCI drug screen database using the ABCG2 functional profile was able to identify novel substrates and transporter-interacting compounds.

Evaluation of current methods used to analyze the expression profiles of ATP-binding cassette transporters yields an improved drug-discovery database

The development of multidrug resistance (MDR) to chemotherapy remains a major challenge in the treatment of cancer. Resistance exists against every effective anticancer drug and can develop by multiple mechanisms. These mechanisms can act individually or synergistically, leading to MDR, in which the cell becomes resistant to a variety of structurally and mechanistically unrelated drugs in addition to the drug initially administered. Although extensive work has been done to characterize MDR mechanisms in vitro, the translation of this knowledge to the clinic has not been successful. Therefore, identifying genes and mechanisms critical to the development of MDR in vivo and establishing a reliable method for analyzing highly homologous genes from small amounts of tissue is fundamental to achieving any significant enhancement in our understanding of MDR mechanisms and could lead to treatments designed to circumvent it. In this study, we use a previously established database that allows the identification of lead compounds in the early stages of drug discovery that are not ATP-binding cassette (ABC) transporter substrates. We believe this can serve as a model for appraising the accuracy and sensitivity of current methods used to analyze the expression profiles of ABC transporters. We found two platforms to be superior methods for the analysis of expression profiles of highly homologous gene superfamilies. This study also led to an improved database by revealing previously unidentified substrates for ABCB1, ABCC1, and ABCG2, transporters that contribute to MDR.

Profiling SLCO and SLC22 genes in the NCI-60 cancer cell lines to identify drug uptake transporters

Molecular and pharmacologic profiling of the NCI-60 cell panel offers the possibility of identifying pathways involved in drug resistance or sensitivity. Of these, decreased uptake of anticancer drugs mediated by efflux transporters represents one of the best studied mechanisms. Previous studies have also shown that uptake transporters can influence cytotoxicity by altering the cellular uptake of anticancer drugs. Using quantitative real-time PCR, we measured the mRNA expression of two solute carrier (SLC) families, the organic cation/zwitterion transporters (SLC22 family) and the organic anion transporters (SLCO family), totaling 23 genes in normal tissues and the NCI-60 cell panel. By correlating the mRNA expression pattern of the SLCO and SLC22 family member gene products with the growth-inhibitory profiles of 1,429 anticancer drugs and drug candidate compounds tested on the NCI-60 cell lines, we identified SLC proteins that are likely to play a dominant role in drug sensitivity. To substantiate some of the SLC-drug pairs for which the SLC member was predicted to be sensitizing, follow-up experiments were performed using engineered and characterized cell lines overexpressing SLC22A4 (OCTN1). As predicted by the statistical correlations, expression of SLC22A4 resulted in increased cellular uptake and heightened sensitivity to mitoxantrone and doxorubicin. Our results indicate that the gene expression database can be used to identify SLCO and SLC22 family members that confer sensitivity to cancer cells.

Improving the prediction accuracy in classification using the combined data sets by ranks of gene expressions

Background

The information from different data sets experimented under different conditions may be inconsistent even though they are performed with the same research objectives. More than that, even when the data sets were generated from the same platform, the data agreement may be affected by the technical variation among the laboratories. In this case, it is necessary to use the combined data set after adjusting the differences between such data sets, for detecting the more reliable information.

Results

The proposed method combines data sets posterior to the discretization of data sets based on the ranks of the gene expression ratios, and the statistical method is applied to the combined data set for predictive gene selection. The efficiency of the proposed method was evaluated using five colon cancer related data sets, which were experimented using cDNA microarrays with different RNA sources, and one experiment utilized oligonucleotide arrays. NCI-60 cell lines data sets were used, which were performed with two different platforms of cDNA microarrays and Affymetrix HU6800 oligonucleotide arrays. The combined data set by the proposed method predicted the test data sets more accurately than the separated data sets did. The biological significant genes were detected from the combined data set, which were missed on the separated data sets.

Conclusion

By transforming gene expressions using ranks, the proposed method is not influenced by systematic bias among chips and normalization method. The method may be especially more useful to find predictive genes from data sets which have different scale in gene expressions.

On the necessity of different statistical treatment for Illumina BeadChip and Affymetrix GeneChip data and its significance for biological interpretation

Background

The original spotted array technology with competitive hybridization of two experimental samples and measuring relative expression levels is increasingly displaced by more accurate platforms that allow determining absolute expression values for a single sample (for example, Affymetrix GeneChip and Illumina BeadChip). Unfortunately, cross-platform comparisons show a disappointingly low concordance between lists of regulated genes between the latter two platforms.

Results

Whereas expression values determined with a single Affymetrix GeneChip represent single measurements, the expression results obtained with Illumina BeadChip are essentially statistical means from several dozens of identical probes. In the case of multiple technical replicates, the data require, therefore, different stistical treatment depending on the platform. The key is the computation of the squared standard deviation within replicates in the case of the Illumina data as weighted mean of the square of the standard deviations of the individual experiments. With an Illumina spike experiment, we demonstrate dramatically improved significance of spiked genes over all relevant concentration ranges. The re-evaluation of two published Illumina datasets (membrane type-1 matrix metalloproteinase expression in mammary epithelial cells by Golubkov et al. Cancer Research (2006) 66, 10460; spermatogenesis in normal and teratozoospermic men, Platts et al. Human Molecular Genetics (2007) 16, 763) significantly identified more biologically relevant genes as transcriptionally regulated targets and, thus, additional biological pathways involved.

Conclusion

The results in this work show that it is important to process Illumina BeadChip data in a modified statistical procedure and to compute the standard deviation in experiments with technical replicates from the standard errors of individual BeadChips. This change leads also to an improved concordance with Affymetrix GeneChip results as the spermatogenesis dataset re-evaluation demonstrates.

Reviewers

This article was reviewed by I. King Jordan, Mark J. Dunning and Shamil Sunyaev.

A disattenuated correlation estimate when variables are measured with error: illustration estimating cross-platform correlations

Previous cross-platform reproducibility studies have compared consistency of intensities as well as consistency of fold changes across different platforms using Pearson's correlation coefficient. In this study, we propose the use of measurement error models for estimating gene-specific correlations. Additionally, gene-specific reliability estimates are shown to be useful in prioritizing clones for sequence verification rather than selecting clones using a simple random sample. The proposed 'disattenuated' correlation may prove useful in a wide variety of studies when both X and Y are measured with error, such as in confirmation studies of microarray gene expression values, wherein more reliable laboratory assays such as real-time polymerase chain reaction are used.

In vitro differential sensitivity of melanomas to phenothiazines is based on the presence of codon 600 BRAF mutation

The panel of 60 human cancer cell lines (the NCI-60) assembled by the National Cancer Institute for anticancer drug discovery is a widely used resource. We previously sequenced 24 cancer genes in those cell lines. Eleven of the genes were found to be mutated in three or more of the lines. Using a pharmacogenomic approach, we analyzed the relationship between drug activity and mutations in those 11 genes (APC, RB1, KRAS, NRAS, BRAF, PIK3CA, PTEN, STK11, MADH4, TP53, and CDKN2A). That analysis identified an association between mutation in BRAF and the antiproliferative potential of phenothiazine compounds. Phenothiazines have been used as antipsychotics and as adjunct antiemetics during cancer chemotherapy and more recently have been reported to have anticancer properties. However, to date, the anticancer mechanism of action of phenothiazines has not been elucidated. To follow up on the initial pharmacologic observations in the NCI-60 screen, we did pharmacologic experiments on 11 of the NCI-60 cell lines and, prospectively, on an additional 24 lines. The studies provide evidence that BRAF mutation (codon 600) in melanoma as opposed to RAS mutation is predictive of an increase in sensitivity to phenothiazines as determined by 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt assay (Wilcoxon P = 0.007). That pattern of increased sensitivity to phenothiazines based on the presence of codon 600 BRAF mutation may be unique to melanomas, as we do not observe it in a panel of colorectal cancers. The findings reported here have potential implications for the use of phenothiazines in the treatment of V600E BRAF mutant melanoma.

Technical variables in high-throughput miRNA expression profiling: much work remains to be done

MicroRNA (miRNA) gene expression profiling has provided important insights into plant and animal biology. However, there has not been ample published work about pitfalls associated with technical parameters in miRNA gene expression profiling. One source of pertinent information about technical variables in gene expression profiling is the separate and more well-established literature regarding mRNA expression profiling. However, many aspects of miRNA biochemistry are unique. For example, the cellular processing and compartmentation of miRNAs, the differential stability of specific miRNAs, and aspects of global miRNA expression regulation require specific consideration. Additional possible sources of systematic bias in miRNA expression studies include the differential impact of pre-analytical variables, substrate specificity of nucleic acid processing enzymes used in labeling and amplification, and issues regarding new miRNA discovery and annotation. We conclude that greater focus on technical parameters is required to bolster the validity, reliability, and cultural credibility of miRNA gene expression profiling studies.

Data mining in genomics

This article reviews important emerging statistical concepts, data mining techniques, and applications that have been recently developed and used for genomic data analysis. First, general background and some critical issues in genomic data mining are summarized. A novel concept of statistical significance is described, the so-called "false discovery rate"-the rate of false-positives among all positive findings-which has been suggested to control the error rate of numerous false-positives in large screening biological data analysis. Two recent statistical testing methods are then introduced: significance analysis of microarray and local pooled error tests. Statistical modeling in genomic data analysis is then presented, such as analysis of variance and heterogeneous error modeling approaches that have been suggested for analyzing microarray data obtained from multiple experimental or biological conditions. Two sections then describe data exploration and discovery tools largely termed as supervised learning and unsupervised learning. The former approaches include several multivariate statistical methods to investigate coexpression patterns of multiple genes, and the latter are the classification methods to discover genomic biomarker signatures for predicting important subclasses of human diseases. The last section briefly summarizes various genomic data mining approaches in biomedical pathway analysis and patient outcome or chemotherapeutic response prediction.

Strategy for encoding and comparison of gene expression signatures

EXALT (EXpression signature AnaLysis Tool) enables comparisons of microarray data across experimental platforms and different laboratories.

EXALT (EXpression signature AnaLysis Tool) is a computational system enabling comparisons of microarray data across experimental platforms and different laboratories . An essential feature of EXALT is a database holding thousands of gene expression signatures extracted from the Gene Expression Omnibus, and encoded in a searchable format. This novel approach to performing global comparisons of shared microarray data may have enormous value when coupled directly with a shared data repository.

Using a seed-network to query multiple large-scale gene expression datasets from the developing retina in order to identify and prioritize experimental targets

Understanding the gene networks that orchestrate the differentiation of retinal progenitors into photoreceptors in the developing retina is important not only due to its therapeutic applications in treating retinal degeneration but also because the developing retina provides an excellent model for studying CNS development. Although several studies have profiled changes in gene expression during normal retinal development, these studies offer at best only a starting point for functional studies focused on a smaller subset of genes. The large number of genes profiled at comparatively few time points makes it extremely difficult to reliably infer gene networks from a gene expression dataset. We describe a novel approach to identify and prioritize from multiple gene expression datasets, a small subset of the genes that are likely to be good candidates for further experimental investigation. We report progress on addressing this problem using a novel approach to querying multiple large-scale expression datasets using a 'seed network' consisting of a small set of genes that are implicated by published studies in rod photoreceptor differentiation. We use the seed network to identify and sort a list of genes whose expression levels are highly correlated with those of multiple seed network genes in at least two of the five gene expression datasets. The fact that several of the genes in this list have been demonstrated, through experimental studies reported in the literature, to be important in rod photoreceptor function provides support for the utility of this approach in prioritizing experimental targets for further experimental investigation. Based on Gene Ontology and KEGG pathway annotations for the list of genes obtained in the context of other information available in the literature, we identified seven genes or groups of genes for possible inclusion in the gene network involved in differentiation of retinal progenitor cells into rod photoreceptors. Our approach to querying multiple gene expression datasets using a seed network constructed from known interactions between specific genes of interest provides a promising strategy for focusing hypothesis-driven experiments using large-scale 'omics' data.

Quantitative methods for genome-scale analysis of in situ hybridization and correlation with microarray data

This study introduces a novel method for standardized relative quantification of colorimetric in situ hybridization signal that enables a large-scale cross-platform expression level comparison of in situ hybridization with two publicly available microarray brain data sources.

With the emergence of genome-wide colorimetric in situ hybridization (ISH) data sets such as the Allen Brain Atlas, it is important to understand the relationship between this gene expression modality and those derived from more quantitative based technologies. This study introduces a novel method for standardized relative quantification of colorimetric ISH signal that enables a large-scale cross-platform expression level comparison of ISH with two publicly available microarray brain data sources.

Application of a correlation correction factor in a microarray cross-platform reproducibility study

BACKGROUND

Recent research examining cross-platform correlation of gene expression intensities has yielded mixed results. In this study, we demonstrate use of a correction factor for estimating cross-platform correlations.

RESULTS

In this paper, three technical replicate microarrays were hybridized to each of three platforms. The three platforms were then analyzed to assess both intra- and cross-platform reproducibility. We present various methods for examining intra-platform reproducibility. We also examine cross-platform reproducibility using Pearson's correlation. Additionally, we previously developed a correction factor for Pearson's correlation which is applicable when X and Y are measured with error. Herein we demonstrate that correcting for measurement error by estimating the "disattenuated" correlation substantially improves cross-platform correlations.

CONCLUSION

When estimating cross-platform correlation, it is essential to thoroughly evaluate intra-platform reproducibility as a first step. In addition, since measurement error is present in microarray gene expression data, methods to correct for attenuation are useful in decreasing the bias in cross-platform correlation estimates.

Profiling studies in ovarian cancer: a review

Ovarian cancer is a heterogeneous disease with respect to histopathology, molecular biology, and clinical outcome. In advanced stages, surgery and chemotherapy result in an approximately 25% overall 5-year survival rate, pointing to a strong need to identify subgroups of patients that may benefit from targeted innovative molecular therapy. This review summarizes: (a) microarray research identifying gene-expression profiles in ovarian cancer; (b) the methodological flaws in the available microarray studies; and (c) applications of pathway analysis to define new molecular subgroups. Microarray technology now permits the analysis of expression levels of thousands of genes. So far seven studies have aimed to identify a genetic profile that can predict survival/clinical outcome and/or response to platinum-based therapy. To date, the clinical evidence of prognostic microarray studies has only reached the level of small retrospective studies, and there are other issues that may explain the nonreproducibility among the reported prognostic profiles, such as overfitting, technical platform differences, and accuracy of measurements. We consider pathway analysis a promising new strategy. The accumulation of small differential expressions within a meaningful molecular regulatory network might lead to a critical threshold level, resulting in ovarian cancer. Microarray technologies have already provided valuable expression data for classifying ovarian cancer and the first clues about which molecular changes in ovarian cancer could be exploited in new treatment strategies. Further improvements in technology as well as in study design, combined with pathway analysis, will allow us to detect even more subtle tumor expression differences among subgroups of ovarian cancer patients. Disclosure of potential conflicts of interest is found at the end of this article.

Use of arrays to investigate the contribution of ATP-binding cassette transporters to drug resistance in cancer chemotherapy and prediction of chemosensitivity

Multidrug resistance (MDR) is a major problem in cancer chemotherapy. One of the best known mechanisms of MDR is the elevated expression of ATP-binding cassette (ABC) transporters. While some members of human ABC transporters have been shown to cause drug resistance with elevated expression, it is not yet known whether the over-expression of other members could also contribute to drug resistance in many model cancer cell lines and clinics. The recent development of microarrays and quantitative PCR arrays for expression profiling analysis of ABC transporters has helped address these issues. In this article, various arrays with limited or full list of ABC transporter genes and their use in identifying ABC transporter genes in drug resistance and chemo-sensitivity prediction will be reviewed.

Switch to an invasive growth phase in melanoma is associated with tenascin-C, fibronectin, and procollagen-I forming specific channel structures for invasion

Malignant melanomas are characterized by their high propensity to invade and metastasize, but the molecular mechanisms of these traits have remained elusive. Our DNA microarray analyses of benign nevi and melanoma tissue specimens revealed that the genes encoding extracellular matrix proteins tenascin-C (TN-C), fibronectin (FN), and procollagen-I (PCOL-I) are highly upregulated in invasive and metastatic melanomas. The expression and distribution of these proteins were further studied by immunohistochemistry in benign nevi, radially and vertically growing melanomas, sentinel node micrometastases, and macrometastases. TN-C was increased in all invasive tumours and metastases, especially at invasion fronts, but not in benign nevi or non-invasive melanomas. Significantly, the intensity of TN-C staining correlated with metastasis to sentinel lymph nodes, better than tumour thickness (Breslow). Moreover, TN-C, FN, and PCOL-I appeared to co-localize in the tumours and form tubular meshworks and channels ensheathing the melanoma cells. Our data suggest that melanoma invasion is associated with the formation of special channel-like structures, providing a new concept, structured tumour cell spreading. Altogether, these data provide potential new prognostic markers and therapeutic targets/strategies for preventing melanoma dissemination.

A latent variable approach for meta-analysis of gene expression data from multiple microarray experiments

BACKGROUND

With the explosion in data generated using microarray technology by different investigators working on similar experiments, it is of interest to combine results across multiple studies.

RESULTS

In this article, we describe a general probabilistic framework for combining high-throughput genomic data from several related microarray experiments using mixture models. A key feature of the model is the use of latent variables that represent quantities that can be combined across diverse platforms. We consider two methods for estimation of an index termed the probability of expression (POE). The first, reported in previous work by the authors, involves Markov Chain Monte Carlo (MCMC) techniques. The second method is a faster algorithm based on the expectation-maximization (EM) algorithm. The methods are illustrated with application to a meta-analysis of datasets for metastatic cancer.

CONCLUSION

The statistical methods described in the paper are available as an R package, metaArray 1.8.1, which is at Bioconductor, whose URL is http://www.bioconductor.org/.

A Marfan syndrome gene expression phenotype in cultured skin fibroblasts

BACKGROUND

Marfan syndrome (MFS) is a heritable connective tissue disorder caused by mutations in the fibrillin-1 gene. This syndrome constitutes a significant identifiable subtype of aortic aneurysmal disease, accounting for over 5% of ascending and thoracic aortic aneurysms.

RESULTS

We used spotted membrane DNA macroarrays to identify genes whose altered expression levels may contribute to the phenotype of the disease. Our analysis of 4132 genes identified a subset with significant expression differences between skin fibroblast cultures from unaffected controls versus cultures from affected individuals with known fibrillin-1 mutations. Subsequently, 10 genes were chosen for validation by quantitative RT-PCR.

CONCLUSION

Differential expression of many of the validated genes was associated with MFS samples when an additional group of unaffected and MFS affected subjects were analyzed (p-value < 3 x 10-6 under the null hypothesis that expression levels in cultured fibroblasts are unaffected by MFS status). An unexpected observation was the range of individual gene expression. In unaffected control subjects, expression ranges exceeding 10 fold were seen in many of the genes selected for qRT-PCR validation. The variation in expression in the MFS affected subjects was even greater.

DNA microarrays in the clinic: how soon, how extensively?

Although DNA microarrays are now widely used in research settings, they have been slow to penetrate clinical practice in spite of their apparent advantages. This is due to the very different requirements for a clinical test in contrast to a research tool, and to a strict necessity for demonstrated clinical utility. There is a clear differentiation between two types of DNA array tests: "genomic" diagnostics, developed to ascertain the presence or absence of mutations, deletions or duplications, and for which clinical evidence is already established, and tests using expression profiling for prognosis or predictive purposes, in which case the clinical correlate must be proven. Most array diagnostics currently used belong, understandably, to the "genomic" variety. It is to be expected that future improvements in tailored technology, as well as a logical trend towards measuring an ever-increasing number of parameters, will ensure an important diagnostic role for DNA arrays in the coming decade.

Review of the literature examining the correlation among DNA microarray technologies

DNA microarray technologies are used in a variety of biological disciplines. The diversity of platforms and analytical methods employed has raised concerns over the reliability, reproducibility and correlation of data produced across the different approaches. Initial investigations (years 2000-2003) found discrepancies in the gene expression measures produced by different microarray technologies. Increasing knowledge and control of the factors that result in poor correlation among the technologies has led to much higher levels of correlation among more recent publications (years 2004 to present). Here, we review the studies examining the correlation among microarray technologies. We find that with improvements in the technology (optimization and standardization of methods, including data analysis) and annotation, analysis across platforms yields highly correlated and reproducible results. We suggest several key factors that should be controlled in comparing across technologies, and are good microarray practice in general.

A survey of methods for classification of gene expression data using evolutionary algorithms

The rapid increase in the quantity of available biologic data over the last decade, brought about by the introduction of massively parallel methods for gene expression measurements, has highlighted the need for more efficient computational techniques for analysis. This paper reviews the use of evolutionary algorithms (EAs) in connection with classification based on gene expression data matrices. Brief introductions to data classification methods and EAs are given, followed by a survey of studies dealing with the application of evolutionary algorithms to various (cancer related) data sets. The general conclusion, based on the published results surveyed here, is that EAs may constitute an efficient method for optimal gene selection, and can also help in reducing the size (number of features used) of classifiers. In many cases, the classification accuracy obtained using EAs, often in conjunction with other methods, represents a significant improvement over results obtained without the use of EAs. However, long-term, independent clinical follow-up studies will be essential to validate prognostic markers identified by the use of EA-based methods.

Transcript and protein expression profiles of the NCI-60 cancer cell panel: an integromic microarray study

To evaluate the utility of transcript profiling for prediction of protein expression levels, we compared profiles across the NCI-60 cancer cell panel, which represents nine tissues of origin. For that analysis, we present here two new NCI-60 transcript profile data sets (A based on Affymetrix HG-U95 and HG-U133A chips; Affymetrix, Santa Clara, CA) and one new protein profile data set (based on reverse-phase protein lysate arrays). The data sets are available online at http://discover.nci.nih.gov in the CellMiner program package. Using the new transcript data in combination with our previously published cDNA array and Affymetrix HU6800 data sets, we first developed a "consensus set" of transcript profiles based on the four different microarray platforms. Using that set, we found that 65% of the genes showed statistically significant transcript-protein correlation, and the correlations were generally higher than those reported previously for panels of mammalian cells. Using the predictive analysis of microarray nearest shrunken centroid algorithm for functional prediction of tissue of origin, we then found that (a) the consensus mRNA set did better than did data from any of the individual mRNA platforms and (b) the protein data seemed to do somewhat better (P = 0.027) on a gene-for-gene basis in this particular study than did the consensus mRNA data, but both did well. Analysis based on the Gene Ontology showed protein levels of structure-related genes to be well predicted by mRNA levels (mean r = 0.71). Because the transcript-based technologies are more mature and are currently able to assess larger numbers of genes at one time, they continue to be useful, even when the ultimate aim is information about proteins.

Gene-expression signatures in oncology diagnostics

The advent of DNA microarray technologies has enabled the development of gene expression signatures that can be used for prognostic and predictive purposes. This new information can change the paradigm of how medicine is practiced, coupling physical examination, pathology and clinical tests with new molecular information. However, many unanswered questions regarding sample acquisition, platform development, signature validation and clinical trial design will need to be addressed before this new medical content will have an impact on the clinical setting. This article will examine some of these issues in greater detail, focusing on tissue type, platform comparison, biospecimen collection and signature validation.

Defining the gene expression signature of rhabdomyosarcoma by meta-analysis

Background

Rhabdomyosarcoma is a highly malignant soft tissue sarcoma in childhood and arises as a consequence of regulatory disruption of the growth and differentiation pathways of myogenic precursor cells. The pathogenic pathways involved in this tumor are mostly unknown and therefore a better characterization of RMS gene expression profile would represent a considerable advance. The availability of publicly available gene expression datasets have opened up new challenges especially for the integration of data generated by different research groups and different array platforms with the purpose of obtaining new insights on the biological process investigated.

Results

In this work we performed a meta-analysis on four microarray and two SAGE datasets of gene expression data on RMS in order to evaluate the degree of agreement of the biological results obtained by these different studies and to identify common regulatory pathways that could be responsible of tumor growth. Regulatory pathways and biological processes significantly enriched has been investigated and a list of differentially meta-profiles have been identified as possible candidate of aggressiveness of RMS.

Conclusion

Our results point to a general down regulation of the energy production pathways, suggesting a hypoxic physiology for RMS cells. This result agrees with the high malignancy of RMS and with its resistance to most of the therapeutic treatments. In this context, different isoforms of the ANT gene have been consistently identified for the first time as differentially expressed in RMS. This gene is involved in anti-apoptotic processes when cells grow in low oxygen conditions. These new insights in the biological processes responsible of RMS growth and development demonstrate the effective advantage of the use of integrated analysis of gene expression studies.

Influence of testosterone on gene expression in the ovariectomized mouse submandibular gland

Androgens exert significant effects on the murine submandibular gland. Our objective in this study was to determine the nature and extent of testosterone regulation of gene expression in the female submandibular gland, and to explore the degree to which this control is the same as in male glands. Ovariectomized female BALB/c mice were treated with placebo- or testosterone-containing hormone pellets for 14 d. Glands were collected and total RNA was isolated. Samples were analyzed for differential expression of mRNA using CodeLink microarrays, and the data were evaluated using genesifter. Testosterone significantly influenced the expression of over 500 genes, and while many (n = 214) of the genes were similarly differentially expressed in androgen-treated males, there were also many that were unique. These findings support our hypotheses that testosterone extensively influences gene expression in the female submandibular gland, and that the nature of this influence is variable between sexes.

Integrating data on DNA copy number with gene expression levels and drug sensitivities in the NCI-60 cell line panel

Chromosome rearrangement, a hallmark of cancer, has profound effects on carcinogenesis and tumor phenotype. We used a panel of 60 human cancer cell lines (the NCI-60) as a model system to identify relationships among DNA copy number, mRNA expression level, and drug sensitivity. For each of 64 cancer-relevant genes, we calculated all 4,096 possible Pearson's correlation coefficients relating DNA copy number (assessed by comparative genomic hybridization using bacterial artificial chromosome microarrays) and mRNA expression level (determined using both cDNA and Affymetrix oligonucleotide microarrays). The analysis identified an association of ERBB2 overexpression with 3p copy number, a finding supported by data from human tumors and a mouse model of ERBB2-induced carcinogenesis. When we examined the correlation between DNA copy number for all 353 unique loci on the bacterial artificial chromosome microarray and drug sensitivity for 118 drugs with putatively known mechanisms of action, we found a striking negative correlation (-0.983; 95% bootstrap confidence interval, -0.999 to -0.899) between activity of the enzyme drug L-asparaginase and DNA copy number of genes near asparagine synthetase in the ovarian cancer cells. Previous analysis of drug sensitivity and mRNA expression had suggested an inverse relationship between mRNA levels of asparagine synthetase and L-asparaginase sensitivity in the NCI-60. The concordance of pharmacogenomic findings at the DNA and mRNA levels strongly suggests further study of L-asparaginase for possible treatment of a low-synthetase subset of clinical ovarian cancers. The DNA copy number database presented here will enable other investigators to explore DNA transcript-drug relationships in their own domains of research focus.

A rapid method for microarray cross platform comparisons using gene expression signatures

Microarray technology has become highly valuable for identifying complex changes in global gene expression patterns. The inevitable use of a variety of different platforms has compounded the difficulty of effectively comparing data between projects, laboratories, and public access databases. The need for consistent, believable results across platforms is fundamental and methods for comparing results across platforms should be as straightforward as possible. We present the results of a study comparing three major, commercially available, microarray platforms (Affymetrix, Agilent, and Illumina). Concordance estimates between platforms was based on mapping of probes to Human Gene Organization (HUGO) gene names. Appropriate data normalization procedures were applied to each dataset followed by the generation of lists of regulated genes using a common significance threshold for all three platforms. As expected, concordance measured by directly comparing gene lists was relatively low (an average 22.8% for all platforms across all possible comparisons). However, when statistical tests (gene set enrichment analysis--GSEA, parametric analysis of gene enrichment--PAGE) which align gene lists with continuous measures of differential gene expression were applied to the cross platform datasets using significant gene lists to poll entire datasets, the relatedness of the results from all three platforms was specific, obvious, and profound.

A sequence-oriented comparison of gene expression measurements across different hybridization-based technologies

Over the last decade, gene expression microarrays have had a profound impact on biomedical research. The diversity of platforms and analytical methods available to researchers have made the comparison of data from multiple platforms challenging. In this study, we describe a framework for comparisons across platforms and laboratories. We have attempted to include nearly all the available commercial and 'in-house' platforms. Using probe sequences matched at the exon level improved consistency of measurements across the different microarray platforms compared to annotation-based matches. Generally, consistency was good for highly expressed genes, and variable for genes with lower expression values as confirmed by quantitative real-time (QRT)-PCR. Concordance of measurements was higher between laboratories on the same platform than across platforms. We demonstrate that, after stringent preprocessing, commercial arrays were more consistent than in-house arrays, and by most measures, one-dye platforms were more consistent than two-dye platforms.