Van Eyk, Dunn - Proteomic and Genomic Analysis of Cardiovascular Disease - 2003 (522919), страница 39
Текст из файла (страница 39)
Biophys. Res. Commun. 1991, 177, 867–873.Sambrook, J., Fritsch, E. F., Maniatis,T., Molecular Cloning, a laboratory manual. Cold Spring Harbor LaboratoryPress, 1989.Jiang, L., Tsubakihara, M., Heinke,M. Y., Yao, M., et al., Heart failure andapoptosis: electrophoretic methods support data from micro- and macro-arrays.A critical review of genomics and proteomics. Proteomics 2001, 1, 1481–1488.Inoue, S., Orimo, A., Saito, T., Ikeda,K., et al., A novel RING finger protein,BFP, predominantly expressed in thebrain. Biochem. Biophys.
Res. Commun.1997, 240, 8–14.Marshall, E., Affymetrix Settles Suit,Fixes Mouse Chips. Science 2001, 291,2535.Knight, J., When the chips are down.Nature 2001, 410, 860–861.1251278Application of Biologic Skepticism to Analysisof Expression PhenotypesLawrence Adams, Roger E.
Bumgarner, and Stephen Mark SchwartzIt is all too obvious that expression arrays have been better at producing a 1950’sversion of the Sears catalog than at constructing an interactive 2002 website. Thisreview will suggest criteria for more useful analysis and publication.8.1What Is an Array?As Fig. 8.1 shows, an expression array is simply a list of values representing theamounts of each mRNA species found in a sample.
This concept greatly simplifies the usual red-green display and points out that “arrays” are a simple list ofRNA quantities resulting from any measurement technique. Tab. 8.1 shows sixdifferent RNA detection methods used to quantify large sets of RNA expressionvalues.Finally, although not usually thought of as an “array”, methods derived fromRNA-based representational difference analysis (RNA-RDA) subtraction suppression hybridization (SSH) [1], use selective forms of PCR to eliminate those sequences found in common between two cDNA collections [2].In the mathematical sense, SSH does produce an array of differentially expressed sequences.
This sort of “all or none” expression data provides a compliment to array hybridization methods since the normalization step means thatSSH may even detect low abundance, but differentially expressed, sequences.These genes can then be quantified by real time PCR or used to construct expression arrays [3]. Like SAGE or bend sequencing, SSH is unbiased; however, SSHrequires extensive sequencing to identify the subtracted clones.Proteomic and Genomic Analysis of Cardiovascular Disease.Edited by Jennifer E.
van Eyk, Michael J. DunnCopyright © 2003 WILEY-VCH Verlag GmbH & Co. KGaA, WeinheimISBN: 3-527-30596-31288 Skeptical Analysis of ArraysAlternate views of 2-color array data.A) A traditional display of DNA array data –data presented as a false color image overlayof the Cy3 and Cy5 data channels.
B) Boththe top and bottom blocks show an expandedview of the same region of A. The differencebetween the two figures is the result of manipulation of the color mapping functions –e.g. determining what level of green or red isdisplayed for a given signal intensity in theCy3 and Cy5 channels. Note that it is possibleto present the reader withFig. 8.1dramatically different views of the data simplyby manipulation of the color mapping functions.
C) A linear display of array data. Thetop graph displays the intensity (proportionalto expression level) while the bottom graphdisplays the Cy3/Cy5 ratio for each spotshown in B. The ordered pairs of (intensity,ratio) data contain similar information content to that in either view in B.
However, therepresentation of the data as ordered (intensity, ratio) pairs provides a less subjectiveview than a false color image.8.2Applications of Array Analysis8.2.1Catalogs by FunctionWith a long enough list of differentially expressed genes, it is almost impossiblenot to find 5–10 on which an author can comment. Moreover, these subjectiveevaluations are often biased by predefined functions, such as “inflammation” or“growth”. Imputing function from changes in RNA level is dangerous. Even if oneaccepts the assumptions that regulation expression correlates with function, smallchanges in expression of a group of functionally-related genes may have little todo with how the cell is using the proteins transcribed by these genes.8.2 Applications of Array AnalysisTab.
8.1 How are array data obtained?SAGE [22, 23]Uses concatemers of short sequences to identify genes represented in a collection of mRNA.Does not require predefined setsof genes; can identify previouslyunknown sequences; can identifygenes by multiple 3' exons; istotally unbiased by predefinedideas.Bead-basedsequencing [24]Uses a bead-based sequence to simultaneously provide large numbers of sequences {http://www.lynxgen.de/}.Does not require predefined setsof genes; can identify previouslyunknown sequences; is totallyunbiased by predefined ideas.High throughputsequencing oflibrariesUses PCR-based amplification; onlyminimal sequence is provided; maydetect but not measure frequency oflow abundance sequences; existingdatabase of sequence tags exceedsthe likely number of total gene transcripts (200,000 vs.
35,000).Methodology is new and untested;even at 106 sequences per run, number of “hits” may not measure frequency of low abundance sequences.Although sequencing of whole libraries is impractical, screening methodscan eliminate redundancy. For example, PCR Select® {www.clontech.com/products/literature/pdf/brochures/PCR-SelectBR.pdf } is often used to decrease numbers of clones to sequence. Detected sequences can then be array spotted and analyzed for abundance.Identifies new sequences; accuratesequence identification; detectionof splice forms; unbiased; able tomeasure abundance even of raresequences if enough sequencingis done.Expensive; depends on accuracy of library construction and amplification.Sequence analysis This method, also called gene calling, [25–27] combines the use of a largeby massnumber of restriction fragments with a database of patterns generated fromknown transcripts {www.curagen.com/technology/frametech.htm}.
A computer analysis is used to generate the set of array data much as mass spectrometry fragment lengths are used to identify proteins [28].Unbiased; able to measureabundance even of rare sequences.cDNA ArraysExpensive; requires sophisticated information technology; abundance ofrare species very dependent on numbers of fragments and thereforequantity of RNA studied; cannotidentify unknown sequences.Most common method now used.Rapid; able to analyze large numbersof sequences at one time; becominginexpensive; applicable to organismsfor which little or no genomic dataexist.Need to store and validate largenumbers of cDNA clones; cross hybridization problems; hybridizationsignal may not prove target sequence is actually present; rapidlychanging technologies; can only analyze what is on the physical array.1291308 Skeptical Analysis of ArraysTab.
8.1 (continued)OligonucleotideDNA ArraysSeveral different kinds of oligonucleotide arrays are available including: Affymetrix© arrays on which oligonucleotides are synthesized directly on asilicon chip via a photo-lithographic process {www.affymetrix.com}; Agilent© inkjet oligo arrays on which oligos are synthesized by spraying reagents onto a glass surface with an inkjet printing head {www.chem.agilent.com/Scripts/Icol.asp?iInd=2}; Arrays constructed by depositing pre-synthesized oligonucleotides onto glass or membrane surfaces probably will be themajor technology as costs drop {www.operon.com}.Oligomers can be custom designedand theoretically will be valid to provide highly accurate data; differencesin oligomeric hybridizations may beused to detect mutant or splicedforms; because oligomers can besynthesized as needed, there is noneed to store and validate largenumbers of cDNA clones; choice ofsets of good oligomers or heuristiccomparisons done with differentsets of oligomers representing eachgene should minimize cross hybridization.Current costs for the most commontype (Affymetrix© chip) are high.New methods, e.g.
the use of theinkjet printing, should reduce costs;are not applicable to an organismwith little or no available sequencedata. As in cDNA arrays, oligonucleotide arrays can only analyze whatis on the physical array. Moreover,because they are so sequence specific, oligomeric sequences are not asuseful as longer clone sequenceswhen crossing species barriers andmay be more prone to error if thegene has an unsuspected spliceform or mutation.In part, the problem with making functional inferences from gene expressionpatterns is a statistical artifact sometimes called the “Kevin Bacon” game [4, 5].The goal of the game is to relate any actor to Kevin Bacon through a series of costarring links in six or less steps.
Ultimately every actor has some relationship toMr. Bacon, albeit at various levels of removal. Similarly, any gene can be related toalmost any biological state or other gene through a small number of publicationsin which the genes “co-star”. Biologic analysis can be further complicated by looking through the literature for papers that might indicate a “link” between observed differentially expressed genes and the biology of interest.8.2.2Functional GenomicsPerhaps the most aggressive effort at using arrays to infer function has been made inyeast which have only *6,200 genes [6]. Ideker et al.
used a combination of DNAmicroarrays, quantitative proteomics, and databases of known physical interactionsto study the yeast galactose-utilization pathway. Access to all these resources allowedthe investigators to formulate new hypotheses about the regulation of galactose utilization and molecular interactions between this and a variety of other metabolic8.2 Applications of Array Analysispathways. Although such comprehensive data are not available in mammalian systems, the availability of genomic mapping data, in combination with expressiondata, is already of great help in focusing attention on the sets of relevant genes inhuman disease. It is worth noting that expression studies alone cannot prove function. For example, a large and functionally miscellaneous set of genes is rapidly upregulated when serum starved cells are fed.
Some of these “early response genes”have clear functions in the cell cycle; the role of others – e.g. protinases, heat shockprotein, and even tissue factor – may only reflect promoter promiscuity.8.2.3Phenotypic DefinitionsThe potential of arrays for producing phenotypes should be compared with themore traditional approach based on morphology.
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
