Docstoc

Full Text - Perspective

Document Sample
Full Text - Perspective Powered By Docstoc
					                                                                                                                                               Perspective




Perspective
Contributions of the EMERALD project to assessing and
improving microarray data quality
Vidar Beisvåg1, Audrey Kauffmann2, James Malone2, Carole Foy3, Marc Salit4, Heinz Schimmel5,
Erik Bongcam-Rudloff 6, Ulf Landegren6, Helen Parkinson2, Wolfgang Huber2, Alvis Brazma2, Arne K.
Sandvik1,7, and Martin Kuiper8
1
  Department of Cancer Research and Molecular Medicine, Norwegian University of Science and Technology,
Trondheim, Norway, 2 European Molecular Biology Laboratory (EMBL) European Bioinformatics Institute,
Cambridge, UK, 3LGC, Queens Road, Teddington, Middlesex, UK, 4National Institute of Standards and
Technology, Gaithersburg, MD, USA, 5European Commission, Joint Research Centre, Institute for Reference
Materials and Measurements (IRMM), Geel, Belgium, 6Department of Genetics and Pathology and Science for
Life Lab Uppsala, Uppsala University, Uppsala, Sweden, 7Department of Gastroenterology and Hepatology, St.
Olav’s University Hospital, Trondheim, Norway, and 8Department of Biology, Norwegian University of Science and
Technology, Trondheim, Norway
BioTechniques 50:27-31 ( January 2011) doi 10.2144/000113591
Keywords: microarray data; quality metrics; ontology; protocols; standards; external standards; meta-analysis


While minimum information about a microarray experiment (MIAME) standards have helped to increase the
value of the microarray data deposited into public databases like ArrayExpress and Gene Expression Omnibus
(GEO), limited means have been available to assess the quality of this data or to identify the procedures used to
normalize and transform raw data. The EMERALD FP6 Coordination Action was designed to deliver approaches
to assess and enhance the overall quality of microarray data and to disseminate these approaches to the microarray
community through an extensive series of workshops, tutorials, and symposia. Tools were developed for assessing
data quality and used to demonstrate how the removal of poor-quality data could improve the power of statistical
analyses and facilitate analysis of multiple joint microarray data sets. These quality metrics tools have been dis-
seminated through publications and through the software package arrayQualityMetrics. Within the framework
provided by the Ontology of Biomedical Investigations, ontology was developed to describe data transformations,
and software ontology was developed for gene expression analysis software. In addition, the consortium has ad-
vocated for the development and use of external reference standards in microarray hybridizations and created the
Molecular Methods (MolMeth) database, which provides a central source for methods and protocols focusing on
microarray-based technologies.

Over the last 15 years, microarray                       analysis has often been the subject of                     Microarray data produced in specific,
technology has grown into a leading                      criticism. In particular, reproducibility has          small-scale experiments provide a rich
approach for large-scale investigation of                been a concern both within a given platform            source of information. However, when
transcriptomes, genomes, and epigenomes                  and in cross-platform comparisons (6). This            large amounts of data generated from
(1–3). Microarray technologies are contin-               has been a particular issue with “homemade”            various independent experiments are to be
ually improving, with new applications                   spotted arrays. However, recent reports seem           analyzed, data quality and proper exper-
being reported regularly. Although newer                 to agree that the technology can provide               iment annotation (metadata) are critical.
approaches based on massively parallel                   reliable results (7,8), even when performed            This notion triggered the formation of
DNA sequencing are maturing as an alter-                 in high-throughput. This is largely due to             the Microarray Gene Expression and
native to high-throughput technology for                 the increased robustness sample prepa-                 Data Society [MGED; now renamed to
nucleic acid analysis (4), low cost and practi-          ration and labeling, as well as a switch from          the Functional Genomics Data Society
cality will sustain microarrays as important             homemade arrays to commercial platforms                (FGED); www.mged.org]. In 2001,
tools in both basic research and, increas-               provided by companies such as Affymetrix               MGED published guidelines for exper-
ingly, for diagnostic applications (5).                  (Santa Clara, CA, USA), Agilent (Santa                 iment descriptions (minimum information
    The quality of the data produced from                Clara, CA, USA), and Illumina (San Diego,              about a microarray experiment; MIAME)
microarray technology for transcriptome                  CA, USA).                                              (9) and proposed a structured data exchange

Vol. 50 | No. 1 | 2011                                                       27                                                    www.BioTechniques.com
Perspective


format (MAGE-TAB) (10). This work has                      a focus on data content. EMERALD was                    beyond what is possible in a standard
subsequently served as a model to develop                  designed to help improve data quality,                  workflow. ArrayExpress also uses both
guidelines and standards for many other high-              both by supporting ongoing initiatives of               of these packages in a pipeline to identify
throughput genomics technologies (11). What                MGED and the External RNA Controls                      high-quality data for integration with
has made MIAME successful is the principle                 Consortium (ERCC) (16), and through                     European Bioinformatics Institute (EBI)
that data supporting a scientific analysis must            dedicated activities described herein. An               databases, such as the ArrayExpress Gene
be made available in a way that makes these                overview of the specific results of these               Expression Atlas (19).
data usable for others (12). Currently, data               activities is summarized in Table 1.
from more than 15,000 different microarray                     A fundamental part of EMERALD
studies have been deposited in MIAME-                      has been the development of specific QMs                QA and significance
compliant public repositories such as ArrayEx-             and the generation of publicly available
press (13), Gene Expression Omnibus (GEO)                  software implementing these metrics (17).               of results
(14), and the Center for Information Biology               This development was expected to serve                  We extensively used the QMs approach
Gene Expression Database (CIBEX) (15).                     multiple purposes: (i) a QM tool could offer            ourselves to demonstrate the benefits of
    Microarray data has become an important                a first line of defense against the submission          pruning larger data sets and removing
resource for data-driven analysis, and metadata            of poor-quality data or retrieval of such data          poor-quality data before further analysis.
and data quality in public microarray data                 from common repositories. Core facil-                   One recent paper (20) addressed the
repositories are the major determinants for                ities and microarray laboratories can use               possibility that important genes can be
success in information extraction and model                QM software to screen their new results                 missed in a statistical analysis if a data set
building. This is of particular importance                 and diagnose problematic array data sets                contains poor-quality arrays. We used the
for systems biology approaches, in which                   that could then either be withheld from                 QM software to identify outlier arrays (i.e.,
microarray data are used to derive models                  submission or replaced by better quality                arrays that contribute data of very different
of biological processes. Whereas the focus                 data. (ii) For data already submitted, a                and therefore presumably low quality),
of MGED has been predominantly on data                     QM tool would allow users contemplating                 thereby compromising the statistical and
context and the quality of metadata, the                   a meta-analysis of public data to consider              biological significance of the analysis.
focus of the EMERALD (Empowering the                       only data of sufficient quality. The array-             The removal of such outlier arrays could
Microarray-based European Research Area                    QualityMetrics Bioconductor package                     significantly enhance the sensitivity of the
to Take a Lead in Development and Exploi-                  (www.bioconductor.org/packages/release/                 analysis, by yielding improved statistical
tation) project has been on data content and               bioc/html/arrayQualityMetrics.html) was                 and biological significance. We note that
the quality of the quantitative data produced              created to fulfill this aim. This QM tool               array data can appear to be outliers because
by the technologies. The EMERALD project                   integrates various existing approaches to               of a genuine biological property of a sample
(www.microarray-quality.org), funded by                    microarray data quality assessment and,                 or peculiarities of the protocol (e.g., for
the European Union (EU) 6th Framework                      in some aspects, has developed them                     reasons that are beyond the reach of an
Programme, was established with three                      further. It recognizes different microarray             automated analysis). Therefore, due to the
specific aims: (i) to help improve microarray              platforms, including Affymetrix, Agilent,               risk of removing data that are of interest to
data quality, by (ii) establishing a quality               Illumina, and homemade two-color arrays,                the analysis, fully automated, unsupervised
metric (QM) tool able to measure quality,                  and computes general and platform-specific              outlier removal is currently not advisable.
and (iii) to build a normalization and trans-              QMs. The tool produces a comprehensive                  Instead, we recommend that automatically
formation ontology to archive methods of data              report reflecting individual array quality,             identified outlier candidates be manually
preprocessing. Toward those objectives, the                including both relative measures of quality             examined before a decision is made. The
EMERALD consortium has worked closely                      within the data set and absolute metrics.               comprehensive report with the graphical
with MGED, and in this article, we describe                To facilitate meta-analyses of public data,             display of results generated by arrayQuali-
and discuss the results of the project.                    the ArrayExpress Bioconductor package                   tyMetrics helps users understand whether
                                                           has also been developed (18). This package              a particular array should be considered an
                                                           provides a bridge from the ArrayExpress                 outlier. Manual inspection can also provide
Quality metrics                                            repository to the R/Bioconductor data                   useful feedback to improve experimental
The EMERALD Coordination Action                            analysis environment, allowing users to                 protocols.
has continued the process toward                           perform a wide variety of customized,                       The results and experiences from
standardization of microarray data, with                   experiment-specific analyses of data quality            transcriptome microarray quality assurance

Table 1. Overview of the results of the EMERALD project.

 Results                                             Available at

  arrayQualityMetrics software                        www.bioconductor.org/packages/stats/bioc/arrayQualityMetrics.html

  ArrayExpress Bioconductor package                   www.bioconductor.org/packages/release/bioc/html/ArrayExpress.html

  Experimental Factor Ontology                        www.ebi.ac.uk/efo

  Gene expression software ontology                   www.ebi.ac.uk/efo/swo

  The Molecular Methods Database                      www.molmeth.org

  Project web site                                    www.microarray-quality.org




Vol. 50 | No. 1 | 2011                                                             28                                                   www.BioTechniques.com
                                                                                                                                   Perspective


(QA) and quality control (QC) will create        concepts that can be computationally           We have, therefore additionally developed
an example for emerging applications of          processed. In certain languages, such as the   an ontology of gene expression software
other high-throughput molecular technol-         W3C-approved Web Ontology Language             that contains details of commonly used
ogies, such as microarray-based protein          (OWL) (22), these rules can be used to         software such as BioConductor packages
analyses. Initial studies with the QM            check that models are consistent (i.e., that   (www.ebi.ac.uk/efo/swo), the algorithms
software on protein microarrays have shown       contradictory statements are not made) and     implemented, and the purposes they can be
that it can detect outlier data similar to its   also aid interoperability by standardizing     used for. This has been shown to be useful
application on transcriptome microarrays         syntax for creating ontologies.                for text mining full-text articles and experi-
(unpublished observations), because many            To describe the types of data transfor-     mental records in ArrayExpress.
of the same metrics (reproducibility of          mations used in gene expression analysis in
replicate measurements, spatial distribution     an ontology, we used a three-step approach.
of the signal, and dynamic range across          First, we collected use cases that were        External standards
replicates) apply also to protein arrays.        initially used to identify requirements        It is well-established that the use of common
We have subsequently demonstrated the            and, later, to test the developed ontology.    external standards (e.g., spikes or reference
utility of the developed microarray QMs          Second, we considered the sorts of domain      RNAs) is helpful to standardize and evaluate
approach in a systems biology application        concepts and relations that ideally would be   experimental results. At the start of the
by building a global map of human gene           described in the ontology and that could       EMERALD project, multiple independent
expression from a compendium of gene             fulfil these use cases. Third, we integrated   approaches were being proposed to develop
expression data generated by 163 different       our efforts with an existing framework,        such standards. These included platform-
laboratories (21). Over 9000 raw data files      the Ontology of Biomedical Investiga-          specific controls developed by individual
produced on the Affymetrix U133A human           tions (OBI; http://obi-ontology.org/page/      array manufacturers and more generic
gene expression array were collected from        Consortium), which is a community-             efforts to develop reference methods and
the GEO (14) and ArrayExpress (13) reposi-       driven ontology built by an international      materials by the grassroots array community,
tories. An integrated analysis of this massive   consortium of over 30 groups, so that it       metrology institutes, regulatory agencies,
microarray data set compiled from many           may increase interoperability with other       and technology providers. Examples of these
different laboratories was used to reveal the    ongoing community efforts.                     initiatives included the Microarray Controls
overall structure of gene expression space,         The value of the data transformation        Consortium (MAQC) (24), the External
after a careful process to remove outlier        ontology is further increased when it is       RNA Controls Consortium (ERCC), the
arrays and imposing stringent criteria for       integrated within the wider context of         Clinical Laboratory Genomic and Genetic
data quality.                                    OBI. OBI has the scope of describing the       Standards (CLGGS) consortium, the
                                                 elements of a biomedical investigation         Clinical and Laboratory Standards Institute
Representing semantically                        including protocols, instruments, and          (CSLI), and array standardization initiatives
                                                 roles of participants in studies. When         under the UK Measurements for Biotech-
rich data transformations                        we couple descriptions of the data trans-      nology program (www.mfbprog.org.uk). A
Whereas our development of QM software           formation parts of experiments with the        key role of EMERALD was to survey, liaise
reflects an extension of the original MGED       broader scope of OBI, we are able to more      with, and assess the various approaches, and
objectives toward data content, we also          richly describe the entire experiment (with    advocate the use of consistent standards by
extended the original focus of MGED by           regard to, for example, instrumentation        the microarray community. During the
enabling the upload of metadata. Funda-          used, specific assaying processes, the input   lifetime of EMERALD, a single consortium
mental to the reproducibility of exper-          material, and the organization or individual   (ERCC) dedicated to developing microarray
imental protocols is the clarity of the          performing a study). This effort represents    reference standards emerged as the major
documentation of the experiments. This           the first substantial cross-community          developer of materials in this field. We
requires the use of a language that is           attempt to support the annotation of the       practically assessed the potential benefits
precise, explicit, unambiguous, and under-       experimental context of biomedical data,       of using the ERCC reference standards,
standable for the scientific community. The      and is an important achievement within         and in several workshops, we disseminated
use of controlled vocabularies can go some       the microarray and wider bioinformatics        the work of the ERCC and EMERALD
way toward achieving this by providing           community. The ontology is already             in developing and evaluating the materials.
a restricted terminology that defines            deployed as part of the Experimental           The ERCC is developing a panel of RNA
important aspects of a given domain or           Factor Ontology (EFO) (23) in the Gene         sequences (16) to serve as a certified reference
application. However, such vocabularies          Expression Atlas, in the Immune Epitope        material (RM; SRM-2374) in the form of 96
lack the ability to formally relate concepts     Database (www.immuneepitope.org),              sequence-verified plasmids containing the
to one another, sometimes resulting in           and in Integrative Tools for Protozoan         sequences. These ERCC panel sequences
an ad hoc “bag of words.” Ontologies             Parasite Research (www.bioontology.            were selected for their cross-platform perfor-
have become an important method for              org/node/604). In the near future, it is       mance and data consistency from an initial
agreeing on cross-domain concepts and for        expected that OBI will become an Open          library of candidate sequences donated
describing experimental processes and data       Biological and Biomedical Ontologies           by consortium members. A publication
and are being implemented in interfaces          (OBO) Foundry ontology; this certificate       outlining platform performance and data
such as the ArrayExpress archive to provide      of quality will thereby promote OBI as a       consistency is currently in preparation.
queries based on technology, software            community standard.
used, and statistical analysis methods (19).        Besides describing the protocols, it is     Certification of the ERCC clones
Ontologies offer the advantage of modeling       also important to describe the software        Normally, RMs must comply with the
explicit relations among concepts, such as       used to generate the data, and these terms     quality criteria cited in International
subclass or part-of, and can contain rules,      are more commonly used in papers than          Organization for Standardization (ISO)
in the form of axioms, about the use of          detailed descriptions of analysis processes.   15194:2002 and ISO Guide 34:2009 (e.g.,

Vol. 50 | No. 1 | 2011                                           29                                                   www.BioTechniques.com
Perspective


characterization, stability, homogeneity, and      will be necessary for producers of RM or         has been downloaded by 2841 distinct
commutability assessment). However, ISO            microarray end-users to express RNA from         internet addresses during the time
15194:2002 only fully applies to RMs that          the clones which can then be spiked into         from August 2009 to July 2010 (http://
possess quantifiable values of either a differ-    microarray experiments. To function as           bioconductor.org/packages/stats/bioc/
ential or rational quantity, which also holds      a certified RM, the RNA produced from            arrayQualityMetrics.html), and it has
true for the revised ISO 15194:2009 standard,      the clones will need to be produced and          been adopted by academic and commercial
but where nominal and ordinal property-            certified in accordance with ISO guide-          microarray core facilities for their report
related aspects have been further clarified.       lines as discussed; however, the materials       generation.
As a result, new criteria for the quality review   will also need to be certified for quantity.
of nucleic acid as RMs are being established,      Evaluation of in vitro transcribed RNA in
and under the ISO Committee on Reference           terms of size, purity, and concentration—        Future perspectives
Materials (REMCO), a working group has             particularly with respect to batch-to-batch      Several studies have shown that cross-
been set up to develop a standard on require-      variation—is essential for the generation of     platform comparisons are more consistent
ments for RMs for qualitative analysis,            robust and reproducible data.                    when microarrays and protocols developed
including nominal properties. Recently,                                                             by commercial companies are used. In
additional guidance has been made available                                                         general, users of those arrays are more
for managing nominal properties, such as           MolMeth database                                 likely to adopt standardized protocols
those generated by sequence base calls (25).       As a further aid to establishing stable          than users of arrays manufactured by
    Recommendations for certifying                 protocols for the production of microarray       core facilities. This will eliminate a signif-
materials based on sequence include: (i)           data, the EMER ALD project has                   icant source of variation across experi-
quality-scored bidirectional sequencing, (ii)      contributed to the development of the            ments. The results and experiences from
verification by alternative sequence assay and     Molecular Methods (MolMeth) database             transcriptome microarray QA/QC will
independent laboratory, and (iii) expression       (www.molmeth.org), which provides the            further create a cornerstone for systems
of uncertainty as the probability of a             research community with an up-to-date            biology–based life science and also be of
miscalled base. In response to the last point,     source of methods and protocols used in          value as a model for standardizing many
the National Institute of Standards and            molecular biology and molecular medicine.        other emerging high-throughput analyses,
Technology (NIST) has developed a best-            All entries in the MolMeth database are          first for use by the broader research
practice approach that utilizes an ordinal         manually curated, meaning that protocols         community, including systems biology
scale to characterize confidence in regions        are checked against published papers or          and medical research, and later as a base
of sequence. This comprises four orders of         verified by consulting the authors before        for applications in routine diagnostics.
descending confidence: 1, most confident           addition to the database. The database           Examples of new analytical technologies
(all data agree); 2, very confident (ambiguity     complements other efforts to standardize         in need of standardization include the
resolved with second strand); 3, confident         molecular techniques by simplifying the          rapidly expanding scope for image analysis
(ambiguity resolved with judgment); and 4,         documentation of experimental proce-             and the various forms of next-generation
ambiguous (ambiguity unresolved).                  dures. The database has a modular, easily        sequencing. In fact, MIAME guidelines
    All of the 96 ERCC external RNA                searchable structure, and it allows for          have already created an offspring—similar
controls have been analyzed by conventional        convenient time-stamped versioning of            guidelines for experiments in which assays
bidirectional Sanger sequencing and next-          protocols with hyperlinks to resources, such     are performed using high-throughput
generation sequencing using the ABI SOLiD          as commercial products or research publica-      sequencing—the minimum information
(Applied Biosystems, Foster City, CA, USA)         tions. While the database is freely available,   about a high-throughput sequencing
and Illumina GA-IIx platforms. NIST will           the user can choose to password-protect an       experiment (MINSEQE; http://mged.
now distribute the ERCC external RNA               entry to confine it to a smaller group before    org/minseqe). These guidelines have been
controls as standard RM 2374 (NIST SRM             making it public, while corresponding data       closely modeled after MIAME. However,
2374) with an estimate of sequence confi-          are being published.                             while initial microarray experiments
dence at each insert base of all 96 clones.                                                         were focused almost exclusively on gene
    The major array vendors have committed                                                          expression, high-throughput assays have
to including content for these sequences on        Other activities                                 a much broader application range, which
their arrays, and the Clinical and Laboratory      The EMERALD project has provided a               adds extra complexity. It is important that
Standards Institute have published guide-          conduit for any ongoing initiatives aimed        major scientific journals adopt MINSEQE
lines for incorporating external RNA               at improving microarray data quality. The        guidelines soon, if the achievements of data
controls in gene expression assays (26).           consortium has been working together with        openness that were facilitated by MIAME
The guidelines cover issues associated             prominent players in the field, including        are not to be lost.
with the use of external RNA controls as           MGED (now FGED), NIST, ERCC,
a tool for verification of technical perfor-       and critical assessment of microarray data
mance. Areas covered include preparation           analysis (CAMDA) to disseminate the              Acknowledgments
of control transcripts, design of primers          results of microarray standardization and        The EMERALD project was funded by the
and amplicons, quality control, use in final       quality improvement initiatives to the           EU commission and the 6th Framework
experiment, and analysis and interpretation        different microarray technology stake-           Programme. The authors wish to thank Ewa
of data obtained.                                  holders. The web portal at EBI hosts a           Sugajska for her administrative role in the
                                                   web site that presents an overview of these      initial phase of the project, Joaquin Dopazo
Requirements for production of                     initiatives (www.microarray-quality.org)         for linking the EMERALD dissemination
RNA from the ERCC controls                         and serves to disseminate the results and        activities with the CAMDA meetings, and
The current ERCC controls will initially           deliverables of the project. The array-          Astrid Lægreid for her driving role in the
only be released as plasmid clones. It             QualityMetrics software package (17)             inception of the project.

Vol. 50 | No. 1 | 2011                                                 30                                                www.BioTechniques.com
Competing interests                                         center for information biology gene expression
                                                            database. C. R. Biol. 326:1079-1082.
The authors declare no competing                        16. Baker, S.C., S.R . Bauer, R .P. Beyer,            The International Journal of Life Science Methods
interests.                                                  J.D. Brenton, B. Bromley, J. Burrill, H.
                                                            Causton, M.P. Conley, et al. 2005. The
                                                            External RNA controls consortium: a progress
References                                                  report. Nat. Methods 2:731-734.
                                                        17. K auf fmann, A., R . Gentleman, and
 1. Gregory, B.D., J. Yazaki, and J.R. Ecker. 2008.         W. Huber. 2009. arrayQualityMetrics—a
    Utilizing tiling microarrays for whole-genome           bioconductor package for quality assessment
    analysis in plants. Plant J. 53:636-644.                of microarray data. Bioinformatics 25:415-
 2. Gresham, D., M.J. Dunham, and D.                        416.
    Botstein. 2008. Comparing whole genomes



                                                                                                              Publish
                                                        18. K au f f ma n n, A . , T. F. R ay ner, H .
    using DNA microarrays. Nat. Rev. Genet.                 Parkinson, M. Kapushesky, M. Lukk, A.
    9:291-302.                                              Brazma, and W. Huber. 2009. Importing
 3. Hohe i s e l , J . D. 2 0 0 6 . M i c ro a r r ay       ArrayExpress datasets into R/Bioconductor.
    technology: beyond transcript profiling and             Bioinformatics 25:2092-2094.
    genotype analysis. Nat. Rev. Genet. 7:200-          19. Kapushesky, M., I. Emam, E. Holloway,



                                                                                                              Your
    210.                                                    P. Kurnosov, A. Zorin, J. Malone, G. Rustici,
 4. Ansorge, W.J. 20 09. Next-generation                    E. Williams, et al. 2010. Gene Expression Atlas
    DNA sequencing techniques. Nat. Biotechnol.             at the European Bioinformatics Institute.
    25:195-203.                                             Nucleic Acids Res. 38(Database issue):D690-
 5. Wu, L., P.M. Williams, and W. Koch. 2005.               D698.
    Clinical applications of microarray-based           20. Kauffmann, A. and W. Huber. 2010.



                                                                                                              Methods.
    diagnostic tests. BioTechniques 39:S577-                Microarray data quality control improves the
    S582.                                                   detection of differentially expressed genes.
 6. Wilkes, T., H. Laux, and C.A. Foy.                      Genomics 95:138-142.
    2007. Microarray data quality—review of             21. Lukk, M., M. Kapushesky, J. Nikkilä,
    current developments. OMICS 11:1-13.                    H. Parkinson, A. Goncalves, W. Huber, E.
 7. Guo, L., E.K. Lobenhofer, C. Wang, R. Shippy,           Ukkonen, and A. Brazma. 2010. A global map
    S.C. Harris, L. Zhang, N. Mei, T. Chen, et              of human gene expression. Nat. Biotechnol.
    al. 2006. Rat toxicogenomic study reveals               28:322-324.
    analytical consistency across microarray            22. Horrock s, I., P.F. Patel-Sch neider,
    platforms. Nat. Biotechnol. 24:1162-1169.               and F. van Harmelen. 2003. From SHIQ and
 8. Ioannidis, J.P., D.B. A l lison, C. A.                  RDF to OWL: the making of a web ontology
    Ball, I. Coulibaly, X. Cui, A.C. Culhane,               language. J. Web Semant. 1:7-26.
    M. Falchi, C. Furlanello, et al. 2009. Repeat-      23. M a l o n e , J . , E . H o l l o w a y, T.
    ability of published microarray gene expression         Adamusiak, M. Kapushesky, J. Zheng, N.
    analyses. Nat. Genet. 41:149-155.                       Kolesnikov, A. Zhukova, A. Brazma, and H.
 9. Brazma, A., P. Hingamp, J. Quackenbush,                 Parkinson. 2010. Modeling sample variables
    G. Sherlock, P. Spellman, C. Stoeckert, J.              with an experimental factor ontology. Bioin-
    Aach, W. Ansorge, et al. 2001. Minimum
    information about a microarray experiment
                                                            formatics 26:1112-1118.                           BioTechniques has been publishing the
                                                        24. M AQ C C o n s o r t i u m . 2 0 0 6 . T h e
    (MIAME)-toward standards for microarray                 MicroArray Quality Control (MAQC) project         latest methods and techniques for over 27
    data. Nat. Genet. 29:365-371.                           shows inter- and intraplatform reproduc-
10. R ay ner, T.F., P. Rocca-Ser ra, P.T.                                                                     years, reaching a global audience of more
                                                            ibility of gene expression measurements. Nat.
    Spellman, H.C. Causton, A. Farne, E.                    Biotechnol. 24:1151-1161.                         than 80,000 researchers. We welcome new
    Holloway, R.A. Irizarry, J. Liu, et al. 2006. A     25. R o u s s e a u , F. , D. G a n c b e r g , H .
    simple spreadsheet-based, MIAME-supportive              Schimmel, M. Neumaier, A. Bureau, C.              articles describing innovative methods,
    format for microarray data: MAGE-TAB.
    BMC Bioinformatics 7:489.
                                                            Mamotte, R. van Schaik, D. Payne, et al.          substantive modifications to existing
                                                            2009. Considerations for the development of
11. Taylor, C.F., D. Field, S.A. Sansone, J.                a reference method for sequencing of haploid      methods, or innovative applications to new
    Aerts, R. Apweiler, M. Ashburner, C.A. Ball,            DNA—an opinion paper on behalf of the IFCC
    P.A. Binz, et al. 2008. Promoting coherent                                                                models or scientific questions.
                                                            Committee on Molecular Diagnostics. Clin.
    minimum reporting guidelines for biological             Chem. Lab. Med. 47:1343-1350.
    and biomedical investigations: the MIBBI            26. Cl i n ica l a nd Laborator y Sta nd a rds
    project. Nat. Biotechnol. 26:889-896.                   Institute. Use of External RNA Controls in        We are particularly interested in submissions
12. Brazma, A . 2 0 0 9. M i n i mu m i nfor-
    mation about a microarray experiment
                                                            Gene Expression Assays; Approved Guideline.       on methods in the fields of proteomics,
                                                            CLSI Document MM16-A. Vol. 26. [ISBN
    (MIAME)—successes, failures, challenges.                1-56238-617-4]. CLSI, Wayne, PA.                  systems biology, cell culture, and
    ScientificWorldJournal 9:420-423.
13. B r a z m a , A . , H . P a r k i n s o n , U.                                                            epigenetics. We offer a rapid time to first
                                                        Received 6 September 2010; accepted 30
    Sarkans, M. Shojatalab, J. Vilo, N. Abeygu-         November 2010.                                        decision (generally less than 4 weeks from
    nawardena, E. Holloway, M. Kapushesky, et
    al. 2003. ArrayExpress—a public repository                                                                date of submission) and publication within
    for microarray gene expression data at the EBI.     Address correspondence to Martin Kuiper,
    Nucleic Acids Res. 31:68-71.                        Department of Biology, Norwegian University           3 months of acceptance.
14. Edgar, R ., M. Domrachev, and A.E.                  of Science and Technology (NTNU),
    Lash. 2002. Gene Expression Omnibus: NCBI           Høgskoleringen 5, 7491 Trondheim, Norway.
    gene expression and hybridization array data        e-mail: kuiper@nt.ntnu.no                             More info: www.biotechniques.com/
    repository. Nucleic Acids Res. 30:207-210.                                                                morefrombiotechniques/forauthors/
15. I keo, K ., J. Ishi-i, T. Tamura, T.                To purchase reprints of this article, contact:
    Gojobori, and Y. Tateno. 2003. CIBEX:               carmelitag@fosterprinting.com                         submitpapers/
                                                                                                              Submit manuscripts via:
                                                                                                              www.editorialmanager.com/bt
Vol. 50 | No. 1 | 2011                                                     31

				
DOCUMENT INFO