Analysis of cancer metabolism with high-throughput technologies
© Markovets and Herman; licensee BioMed Central Ltd. 2011
Published: 18 October 2011
Recent advances in genomics and proteomics have allowed us to study the nuances of the Warburg effect – a long-standing puzzle in cancer energy metabolism – at an unprecedented level of detail. While modern next-generation sequencing technologies are extremely powerful, the lack of appropriate data analysis tools makes this study difficult. To meet this challenge, we developed a novel application for comparative analysis of gene expression and visualization of RNA-Seq data.
We analyzed two biological samples (normal human brain tissue and human cancer cell lines) with high-energy, metabolic requirements. We calculated digital topology and the copy number of every expressed transcript. We observed subtle but remarkable qualitative and quantitative differences between the citric acid (TCA) cycle and glycolysis pathways. We found that in the first three steps of the TCA cycle, digital expression of aconitase 2 (ACO2) in the brain exceeded both citrate synthase (CS) and isocitrate dehydrogenase 2 (IDH2), while in cancer cells this trend was quite the opposite. In the glycolysis pathway, all genes showed higher expression levels in cancer cell lines; and most notably, digital gene expression of glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and enolase (ENO) were considerably increased when compared to the brain sample.
The variations we observed should affect the rates and quantities of ATP production. We expect that the developed tool will provide insights into the subtleties related to the causality between the Warburg effect and neoplastic transformation. Even though we focused on well-known and extensively studied metabolic pathways, the data analysis and visualization pipeline that we developed is particularly valuable as it is global and pathway-independent.
Profound differences between the metabolic pathways of normal and cancer cells have been known since 1926 when Otto Warburg attributed tumors to dysfunctional mitochondria . Normal eukaryotic cells generate energy in the form of adenosine triphosphate (ATP) through a combination of glycolysis and the TCA pathway. The TCA pathway consists of enzymes encoded by 8 genes and the full cycle generates 36 molecules of ATP. On the other hand, glycolysis is a linear pathway that comprises enzymes encoded by 10 genes and produces only 2 molecules of ATP [2, 3]. Under normal conditions glycolysis preferentially occurs in hypoxic circumstances yielding lactic acid . In many types of cancer, glycolysis is the preferable method for meeting energy demands in cells undergoing uncontrolled growth even in the presence of oxygen . This drastic metabolic shift is essentially the well-known Warburg effect . While enzymes and substrates of the TCA cycle and glycolysis have been extensively studied, the causality between the Warburg effect and cancer remains unknown .
Improvements in ‘-omics’-based technologies, in particular genomics and proteomics, have allowed us to examine at a new level the molecular details of the Warburg effect. For example, it has been found that tumor glycolysis enhances activation of oncogene and loss of tumor suppressor gene activity by stabilization of the hypoxia-inducible factor (HIF), a transcription factor that regulates 9 out of 10 enzymes involved in glycolysis . However, molecular differences between the TCA cycle and glycolysis and the impact of the expression of genes on rates and quantities of ATP production is currently unknown.
Metabolic pathways are regulated at several levels, including mRNA transcription, translation, and protein interaction . We used next- (or second-) generation sequencing technology with RNA-Seq to study the TCA cycle and glycolysis at the transcriptional level of genes encoding their enzymes. This genomics approach is based on the analysis of digital gene expression and is capable of generating millions of short sequences [7, 8]. While the second-generation sequencing technology can provide unprecedented levels of detail, one of its main challenges is the unavailability of intuitive, publicly available universal tools that can retrieve, process and visualize large amounts of generated data in a single package. Hundreds of computational programs for analysis of massive quantities of short reads are already available in the public domain . These programs were designed either for read alignment [10, 11], analysis of alternative splicing and gene expression [12, 13] or visualization [14–16]. Our contribution with the presented work is creation of a package that leverages publicly available programs for next generation sequencing data analysis. This package aligns “reads” (or cDNA sequences), quantifies gene expression (in terms of RNA level) and visualizes results in a simple flow of linear data analysis.
Interpretation of second-generation sequencing data requires expertise in quality control, alignment of reads, quantitation of expression and visualization of massive volumes of data. These steps are commonly embedded in custom-made data analysis pipelines that consist of several steps and typically use some of the aforementioned programs. The data analysis process is performed in a specific order in which output of one operation is used as an input for the next one. Such an approach is highly non-trivial and may still present a challenge for sequencing facilities significantly smaller in size and resources than the National Human Genome Research Institute-funded large genomic centers . To meet this challenge, we have developed Transcriptome Analysis with Circos (TrAC), a novel tool for comparative analysis and visualization of short reads based on Circos . TrAC is a highly customizable RNA-Seq tool applicable for global transcriptome analysis with the additional feature that the visualization step deviates from the “one-gene-at-a-time” paradigm commonly practiced by the popular genome browsers [14, 15]. With TrAC users can visualize whole pathways, cycles within pathways or linear segments of genes in any given configuration. We applied this tool to study energy metabolism in tissues with high-energy requirements such as normal brain and cancer cell lines.
Results and discussion
Core genes of the TCA cycle and glycolysis
mRNA accession number
Copy Number of Transcript
(isocitrate dehydrogenase 2)
(succinate dehydrogenase complex, sub. A)
(malate dehydrogenase 1)
(aldolase A, fructose-bisphosphate)
(triosephosphate isomerase 1)
(phosphoglycerate kinase 1)
(phosphoglycerate mutase 1)
RNA-Seq data processing
In order to provide biological meaning for the reads mapped on the reference and to report the digital expression signal as a scalar value, for each gene we considered both the number of mapped reads and the mapped length in base pairs. To quantify this expression, the ratio between mapped reads and transcribed length was appropriately rescaled with the total number of mapped reads and the total mapped transcriptome size, as previously described . An example of the raw read count is shown in Figure 1A, while estimates of expression of each exon are shown in Figure 1B. Under the assumption of uniform read coverage, this copy number of transcripts corresponds to the amount of every expressed RNA present in the sample and is directly proportional to the RPKM (Reads mapping to the genome Per Kilobase of transcript per Million reads sequenced) value .
The TCA cycle analysis
Visualization of digital expression demonstrated that sequence coverage by transcriptomic tool RNA-Seq is non-uniform but exhibits a highly reproducible pattern across different samples, as shown in Figure 2. The patterns were especially obvious in the case of ACO2 and MDH1 in the TCA cycle. In both samples, the ACO2 and SDHA genes were well covered with reads along the whole length of the transcripts. However, the MDH coverage was biased towards the 5’ end and the middle of the transcript, whereas very few reads mapped to the 3’ end (Figure 2A).
Every gene in the glycolysis pathway showed stronger expression in neoplastic cells than in the brain. The absolute transcript numbers of genes in both samples followed a similar pattern within first five steps in the pathway: the expression level gradually increased. However there were several steps in glycolysis that had very high levels of expression in both types of cells. In particular the digital signal for GAPDH and ENO were considerably higher in both samples when compared to other genes in the pathway. Furthermore, in cancer cell lines the copy numbers of GAPDH and ENO were 3 and 5 times higher than in brain tissue, respectively (Figure 2B).
Our approach revealed both quantitative signals for digital expression and topology of transcript coverage for all analyzed genes. The next generation mRNA sequencing provides an unbiased view of complete transcriptomes. To begin to make sense of transcriptomes represented by large and growing volumes of sequencing data, we need to be able to focus on individual subsets of genes. Topologically speaking, every pathway consists of one of the two major building blocks: a closed loop (such as the TCA cycle) and a linear cascade (such as glycolysis). The tool that we present, Transcriptome Analysis with Circos or TrAC, is our version of an unbiased approach to data analysis because it allows end users to pursue both hypothesis-driven and hypothesis-generating research. In hypothesis- driven analyses, the genes that participate in the major pathways are mainly known. TrAC is also suitable for hypothesis-generating research because an arbitrary list of genes that represent substructures of transcriptional networks can be very efficiently quantitated and visualized. This is a significant advantage over popular genome viewers such as Integrated Genomics Viewer, developed at the Broad Institute, that is constrained to visualize a single gene or a contiguous stretch of DNA on the same chromosome, but by default cannot quantitate gene expression or be used for differential expression. TrAC, therefore, represents a shift in NGS data analysis because it can do both. Moreover it is very flexible about its input, as it simply requires a list of fastq files and a list of genes of interest for analysis.
To appreciate the discovery potential of TrAC, we examined two major, extremely well-studied carbohydrate metabolic pathways for which we still have a very limited understanding of the transcriptional control and consequences of gene activity.
The differences that we observed in ENO expression in cancer cell lines versus brain support the hypothesis that ENO plays a role in pyruvate channeling towards the TCA cycle in mitochondria . On the other hand, with the exception of breast, head and neck, and bone marrow cancers, ENO tends to be overexpressed in a majority of cancers . With the mitochondrial aconitase gene, ACO2, it is quite the opposite as it is generally overexpressed in a small number of cancers such as melanoma and lung cancers . In the P3 prostate cancer cell line, inhibition of ACO2 did not cause a major shift in ATP production, but inhibition of both glycolysis and respiration were necessary to decrease the ATP content . From a bioinformatics standpoint, the ACO2 transcript is of particular interest because a microarray probe on the human Illumina gene arrays cannot be mapped using stringent mapping criteria (MAQC) since the probe has two mismatches relative to the reference. On the other hand, two mismatches on a 75bp-long read in next generation mRNA sequencing experiments are less of a problem because large numbers of reads do not all have to be of superb quality to provide confidence about gene expression.
Biological interpretation of the results presented would require experimental design beyond the scope of this study. In the absence of a carefully designed experiment to study changes in a global gene expression network due to manipulation of the main genes in the TCA cycle and glycolysis, we can attempt to understand the gene activity associated with these two processes through visualization with TrAC. While the objective of the work presented was not to explain the long-standing puzzle of the Warburg effect, the described package, TrAC, represents a viable tool for guiding the analysis. We expect that TrAC will simplify data analysis for non-bioinformaticians (manuscript in preparation). As discussed, TrAC can be scaled up to hundreds of genes involved in metabolic pathways; and we propose it as a bioinformatics tool for a simplified multi-gene rather than single-gene approach in biological problems.
Next-generation sequencing technology was used to study gene expression/activity differences in metabolic pathways between normal and neoplastic cells. For this study we developed and implemented TrAC – Transcriptome Analysis with Circos, a novel computational data analysis and visualization tool. With the TrAC pipeline, we were able to process RNA-Seq data and from large volumes of sequencing data extract meaningful insights into a biological problem. Transcript copy numbers of the main genes involved in carbohydrate metabolism in brain and cancer cells were estimated and different gene expression patterns within two samples were revealed. We expect that this analytical tool will provide further insights into the subtleties related to the causality of the Warburg effect and neoplastic transformation. Our future research will involve translational investigation of the influence of described variations on the rates and qualities of ATP production in normal and neoplastic cells. Although we focused on the genes involved in the main steps of the TCA cycle and glycolysis, the developed approach is global and pathway-independent. TrAC is fully customizable and allows end-users to study any gene expression-related biological question (manuscript, in preparation).
Analysis of energy metabolism was performed on high quality RNA samples of normal brain and cancer cell lines. The brain sample consisted of FirstChoice Human Brain Reference RNA (HBRR by Ambion, cat #AM6050), pooled from multiple donors and several brain regions. The cancer sample was the Universal Human Reference RNA (UHRR by Stratagene), composed of RNA from 10 human cell lines . These samples were sequenced by Illumina using the Genome Analyzer platform for the Sequencing Quality Control project, in a follow up to a large FDA-led Microarray Quality Control (MAQC Consortium) project . In particular, we used 46.8 million HBRR and 50.6 million UHRR in 50 nt-long single-end reads.
A pipeline for analysis and visualization of RNA-Seq data
The analysis of RNA-Seq data consisted of several steps: pre-processing, alignment, post-processing, and visualization. The pre-processing step consists of preparing reads for the alignment. The information content of each unique read was calculated according to the Shannon entropy formula (manuscript in preparation). The rationale for this filtering was our finding that the aligner tends to spend most of the CPU time aligning uninformative reads such as mononucleotide repeats from polyA tails or dinucleotide simple repeats; so filtering out such reads prior to the mapping significantly improved run times. Second, reads were sorted in an alphabetical order that facilitated the alignment process by providing the aligner with efficient data structure. After this initial step in the data analysis, the high information content unique reads were aligned with Bowtie , a freely available, memory- efficient and very fast short-read alignment software package. We aligned against the RefSeq  release downloaded in October of 2010 with a query to the NCBI web site as previously described . We allowed up to two mismatches in the alignment. Mapped reads were saved in sorted SAM format that was used for gene expression quantitation estimates and post-processing visualization. We plotted the pileup vectors of digital expression for each expressed transcript to visualize transcript topology and read coverage. The number of reads and the covered transcript length were further used to estimate the number of transcribed mRNA copies of genes involved in main steps of cell metabolism. These results were finally visualized using the free software package Circos .
In order to study topology of the transcripts, genes, or exons for reads aligned on the RefSeq reference, we generated a pile-vector for each transcript. This vector consisted of an integer count of mapped read coverage at each nucleotide position in a reference. Visualization of pile-vector allowed us to investigate the different features of transcript regions that we were interested in.
where N Gene is a number of mappable reads aligned on the gene’s exons; L Gene is the sum of the gene’s exon lengths; N Transcriptome is the total number of mappable reads in the experiment; L Transcriptome is the length of the transcriptome.
We wish to thank Dr. Helen Benes for comments and suggestions that helped to improve and clarify this manuscript. This work was in part supported by the NIH Grant Number P20 RR-16460 from the IDeA Networks of Biomedical Research Excellence (INBRE) Program of the National Center for Research Resources.
This article has been published as part of BMC Bioinformatics Volume 12 Supplement 10, 2011: Proceedings of the Eighth Annual MCBIOS Conference. Computational Biology and Bioinformatics for a New Decade. The full contents of the supplement are available online at http://www.biomedcentral.com/1471-2105/12?issue=S10.
- WARBURG O: On the origin of cancer cells. Science 1956, 123(3191):309–314. 10.1126/science.123.3191.309View ArticlePubMedGoogle Scholar
- Kim JW, Dang CV: Cancer's molecular sweet tooth and the Warburg effect. Cancer Res 2006, 66(18):8927–8930. 10.1158/0008-5472.CAN-06-1501View ArticlePubMedGoogle Scholar
- Altenberg B, Greulich KO: Genes of glycolysis are ubiquitously overexpressed in 24 cancer classes. Genomics 2004, 84(6):1014–1020. 10.1016/j.ygeno.2004.08.010View ArticlePubMedGoogle Scholar
- Lopez-Lazaro M: The warburg effect: why and how do cancer cells activate glycolysis in the presence of oxygen? Anticancer Agents Med Chem 2008, 8(3):305–312. 10.2174/187152008783961932View ArticlePubMedGoogle Scholar
- Levine AJ, Puzio-Kuter AM: The control of the metabolic switch in cancers by oncogenes and tumor suppressor genes. Science 2010, 330(6009):1340–1344. 10.1126/science.1193494View ArticlePubMedGoogle Scholar
- Hsu PP, Sabatini DM: Cancer cell metabolism: Warburg and beyond. Cell 2008, 134(5):703–707. 10.1016/j.cell.2008.08.021View ArticlePubMedGoogle Scholar
- Bentley DR, Balasubramanian S, Swerdlow HP, Smith GP, Milton J, Brown CG, Hall KP, Evers DJ, Barnes CL, Bignell HR, Boutell JM, Bryant J, Carter RJ, Keira Cheetham R, Cox AJ, Ellis DJ, Flatbush MR, Gormley NA, Humphray SJ, Irving LJ, Karbelashvili MS, Kirk SM, Li H, Liu X, Maisinger KS, Murray LJ, Obradovic B, Ost T, Parkinson ML, Pratt MR, Rasolonjatovo IM, Reed MT, Rigatti R, Rodighiero C, Ross MT, Sabot A, Sankar SV, Scally A, Schroth GP, Smith ME, Smith VP, Spiridou A, Torrance PE, Tzonev SS, Vermaas EH, Walter K, Wu X, Zhang L, Alam MD, Anastasi C, Aniebo IC, Bailey DM, Bancarz IR, Banerjee S, Barbour SG, Baybayan PA, Benoit VA, Benson KF, Bevis C, Black PJ, Boodhun A, Brennan JS, Bridgham JA, Brown RC, Brown AA, Buermann DH, Bundu AA, Burrows JC, Carter NP, Castillo N, Chiara E, Catenazzi M, Chang S, Neil Cooley R, Crake NR, Dada OO, Diakoumakos KD, Dominguez-Fernandez B, Earnshaw DJ, Egbujor UC, Elmore DW, Etchin SS, Ewan MR, Fedurco M, Fraser LJ, Fuentes Fajardo KV, Scott Furey W, George D, Gietzen KJ, Goddard CP, Golda GS, Granieri PA, Green DE, Gustafson DL, Hansen NF, Harnish K, Haudenschild CD, Heyer NI, Hims MM, Ho JT, Horgan AM, Hoschler K, Hurwitz S, Ivanov DV, Johnson MQ, James T, Huw Jones TA, Kang GD, Kerelska TH, Kersey AD, Khrebtukova I, Kindwall AP, Kingsbury Z, Kokko-Gonzales PI, Kumar A, Laurent MA, Lawley CT, Lee SE, Lee X, Liao AK, Loch JA, Lok M, Luo S, Mammen RM, Martin JW, McCauley PG, McNitt P, Mehta P, Moon KW, Mullens JW, Newington T, Ning Z, Ling Ng B, Novo SM, O'Neill MJ, Osborne MA, Osnowski A, Ostadan O, Paraschos LL, Pickering L, Pike AC, Pike AC, Chris Pinkard D, Pliskin DP, Podhasky J, Quijano VJ, Raczy C, Rae VH, Rawlings SR, Chiva Rodriguez A, Roe PM, Rogers J, Rogert Bacigalupo MC, Romanov N, Romieu A, Roth RK, Rourke NJ, Ruediger ST, Rusman E, Sanches-Kuiper RM, Schenker MR, Seoane JM, Shaw RJ, Shiver MK, Short SW, Sizto NL, Sluis JP, Smith MA, Ernest Sohna Sohna J, Spence EJ, Stevens K, Sutton N, Szajkowski L, Tregidgo CL, Turcatti G, Vandevondele S, Verhovsky Y, Virk SM, Wakelin S, Walcott GC, Wang J, Worsley GJ, Yan J, Yau L, Zuerlein M, Rogers J, Mullikin JC, Hurles ME, McCooke NJ, West JS, Oaks FL, Lundberg PL, Klenerman D, Durbin R, Smith AJ: Accurate whole human genome sequencing using reversible terminator chemistry. Nature 2008, 456(7218):53–59. 10.1038/nature07517PubMed CentralView ArticlePubMedGoogle Scholar
- Wang Z, Gerstein M, Snyder M: RNA-Seq: a revolutionary tool for transcriptomics. Nat Rev Genet 2009, 10(1):57–63. 10.1038/nrg2484PubMed CentralView ArticlePubMedGoogle Scholar
- Perkel JM: Sequence Analysis 101: A newbie’s guide to crunching next-generation sequencing data. The Scientist 2011, 25(3):60.Google Scholar
- Langmead B, Trapnell C, Pop M, Salzberg SL: Ultrafast and memory-efficient alignment of short DNA sequences to the human genome. Genome Biol 2009, 10(3):R25. 10.1186/gb-2009-10-3-r25PubMed CentralView ArticlePubMedGoogle Scholar
- Li H, Ruan J, Durbin R: Mapping short DNA sequencing reads and calling variants using mapping quality scores. Genome Res 2008, 18(11):1851–1858. 10.1101/gr.078212.108PubMed CentralView ArticlePubMedGoogle Scholar
- Trapnell C, Pachter L, Salzberg SL: TopHat: discovering splice junctions with RNA-Seq. Bioinformatics 2009, 25(9):1105–1111. 10.1093/bioinformatics/btp120PubMed CentralView ArticlePubMedGoogle Scholar
- Trapnell C, Williams BA, Pertea G, Mortazavi A, Kwan G, van Baren MJ, Salzberg SL, Wold BJ, Pachter L: Transcript assembly and quantification by RNA-Seq reveals unannotated transcripts and isoform switching during cell differentiation. Nat Biotechnol 2010, 28(5):511–515. 10.1038/nbt.1621PubMed CentralView ArticlePubMedGoogle Scholar
- Kent WJ, Sugnet CW, Furey TS, Roskin KM, Pringle TH, Zahler AM, Haussler D: The human genome browser at UCSC. Genome Res 2002, 12(6):996–1006.PubMed CentralView ArticlePubMedGoogle Scholar
- Robinson JT, Thorvaldsdottir H, Winckler W, Guttman M, Lander ES, Getz G, Mesirov JP: Integrative genomics viewer. Nat Biotechnol 2011, 29(1):24–26. 10.1038/nbt.1754PubMed CentralView ArticlePubMedGoogle Scholar
- Krzywinski M, Schein J, Birol I, Connors J, Gascoyne R, Horsman D, Jones SJ, Marra MA: Circos: an information aesthetic for comparative genomics. Genome Res 2009, 19(9):1639–1645. 10.1101/gr.092759.109PubMed CentralView ArticlePubMedGoogle Scholar
- Koboldt DC, Ding L, Mardis ER, Wilson RK: Challenges of sequencing human genomes. Brief Bioinform 2010, 11(5):484–498. 10.1093/bib/bbq016PubMed CentralView ArticlePubMedGoogle Scholar
- Pruitt KD, Maglott DR: RefSeq and LocusLink: NCBI gene-centered resources. Nucleic Acids Res 2001, 29(1):137–140. 10.1093/nar/29.1.137PubMed CentralView ArticlePubMedGoogle Scholar
- Maglott D, Ostell J, Pruitt KD, Tatusova T: Entrez Gene: gene-centered information at NCBI. Nucleic Acids Res 2011, 39(Database issue):D52–7.PubMed CentralView ArticlePubMedGoogle Scholar
- Mortazavi A, Williams BA, McCue K, Schaeffer L, Wold B: Mapping and quantifying mammalian transcriptomes by RNA-Seq. Nat Methods 2008, 5(7):621–628. 10.1038/nmeth.1226View ArticlePubMedGoogle Scholar
- McCammon MT, Epstein CB, Przybyla-Zawislak B, McAlister-Henn L, Butow RA: Global transcription analysis of Krebs tricarboxylic acid cycle mutants reveals an alternating pattern of gene expression and effects on hypoxic and oxidative genes. Mol Biol Cell 2003, 14(3):958–972. 10.1091/mbc.E02-07-0422PubMed CentralView ArticlePubMedGoogle Scholar
- Singh KK, Desouki MM, Franklin RB, Costello LC: Mitochondrial aconitase and citrate metabolism in malignant and nonmalignant human prostate tissues. Mol Cancer 2006, 5: 14. 10.1186/1476-4598-5-14PubMed CentralView ArticlePubMedGoogle Scholar
- Franklin RB, Costello LC: Zinc as an anti-tumor agent in prostate cancer and in other cancers. Arch Biochem Biophys 2007, 463(2):211–217. 10.1016/j.abb.2007.02.033PubMed CentralView ArticlePubMedGoogle Scholar
- Brandina I, Graham J, Lemaitre-Guillier C, Entelis N, Krasheninnikov I, Sweetlove L, Tarassov I, Martin RP: Enolase takes part in a macromolecular complex associated to mitochondria in yeast. Biochim Biophys Acta 2006, 1757(9–10):1217–1228. 10.1016/j.bbabio.2006.07.001View ArticlePubMedGoogle Scholar
- Matheson BK, Adams JL, Zou J, Patel R, Franklin RB: Effect of metabolic inhibitors on ATP and citrate content in PC3 prostate cancer cells. Prostate 2007, 67(11):1211–1218. 10.1002/pros.20617View ArticlePubMedGoogle Scholar
- Novoradovskaya N, Whitfield ML, Basehore LS, Novoradovsky A, Pesich R, Usary J, Karaca M, Wong WK, Aprelikova O, Fero M, Perou CM, Botstein D, Braman J: Universal Reference RNA as a standard for microarray experiments. BMC Genomics 2004, 5(1):20. 10.1186/1471-2164-5-20PubMed CentralView ArticlePubMedGoogle Scholar
- MAQC Consortium, Shi L, Reid LH, Jones WD, Shippy R, Warrington JA, Baker SC, Collins PJ, de Longueville F, Kawasaki ES, Lee KY, Luo Y, Sun YA, Willey JC, Setterquist RA, Fischer GM, Tong W, Dragan YP, Dix DJ, Frueh FW, Goodsaid FM, Herman D, Jensen RV, Johnson CD, Lobenhofer EK, Puri RK, Schrf U, Thierry-Mieg J, Wang C, Wilson M, Wolber PK, Zhang L, Amur S, Bao W, Barbacioru CC, Lucas AB, Bertholet V, Boysen C, Bromley B, Brown D, Brunner A, Canales R, Cao XM, Cebula TA, Chen JJ, Cheng J, Chu TM, Chudin E, Corson J, Corton JC, Croner LJ, Davies C, Davison TS, Delenstarr G, Deng X, Dorris D, Eklund AC, Fan XH, Fang H, Fulmer-Smentek S, Fuscoe JC, Gallagher K, Ge W, Guo L, Guo X, Hager J, Haje PK, Han J, Han T, Harbottle HC, Harris SC, Hatchwell E, Hauser CA, Hester S, Hong H, Hurban P, Jackson SA, Ji H, Knight CR, Kuo WP, LeClerc JE, Levy S, Li QZ, Liu C, Liu Y, Lombardi MJ, Ma Y, Magnuson SR, Maqsodi B, McDaniel T, Mei N, Myklebost O, Ning B, Novoradovskaya N, Orr MS, Osborn TW, Papallo A, Patterson TA, Perkins RG, Peters EH, Peterson R, Philips KL, Pine PS, Pusztai L, Qian F, Ren H, Rosen M, Rosenzweig BA, Samaha RR, Schena M, Schroth GP, Shchegrova S, Smith DD, Staedtler F, Su Z, Sun H, Szallasi Z, Tezak Z, Thierry-Mieg D, Thompson KL, Tikhonova I, Turpaz Y, Vallanat B, Van C, Walker SJ, Wang SJ, Wang Y, Wolfinger R, Wong A, Wu J, Xiao C, Xie Q, Xu J, Yang W, Zhang L, Zhong S, Zong Y, Slikker W Jr: The MicroArray Quality Control (MAQC) project shows inter- and intraplatform reproducibility of gene expression measurements. Nat Biotechnol 2006, 24(9):1151–1161. 10.1038/nbt1239PubMed CentralView ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.