Autocorrelation analysis reveals widespread spatial biases in microarray experiments
© Koren et al; licensee BioMed Central Ltd. 2007
Received: 12 February 2007
Accepted: 12 June 2007
Published: 12 June 2007
DNA microarrays provide the ability to interrogate multiple genes in a single experiment and have revolutionized genomic research. However, the microarray technology suffers from various forms of biases and relatively low reproducibility. A particular source of false data has been described, in which non-random placement of gene probes on the microarray surface is associated with spurious correlations between genes.
In order to assess the prevalence of this effect and better understand its origins, we applied an autocorrelation analysis of the relationship between chromosomal position and expression level to a database of over 2000 individual yeast microarray experiments. We show that at least 60% of these experiments exhibit spurious chromosomal position-dependent gene correlations, which nonetheless appear in a stochastic manner within each experimental dataset. Using computer simulations, we show that large spatial biases caused in the microarray hybridization step and independently of printing procedures can exclusively account for the observed spurious correlations, in contrast to previous suggestions. Our data suggest that such biases may generate more than 15% false data per experiment. Importantly, spatial biases are expected to occur regardless of microarray design and over a wide range of microarray platforms, organisms and experimental procedures.
Spatial biases comprise a major source of noise in microarray studies; revision of routine experimental practices and normalizations to account for these biases may significantly and comprehensively improve the quality of new as well as existing DNA microarray data.
With the availability of complete genome sequences, the ability to probe multiple genes in a single experiment using DNA microarrays provides an unprecedented tool for genomic research. Accordingly, tens of thousands of microarray experiments have been conducted to monitor changes in gene expression, identify genome-wide protein binding sites, characterize genetic variability and more. Overall, the microarray technology is of ever-increasing usefulness for multiple sorts of biological inquiries.
DNA microarrays are composed of numerous probes that usually interrogate a complete genome. The different sequence-specific probes are arrayed on a single surface either by in-situ oligonucleotide synthesis, or by spotting gene-specific nucleic acid fragments organized in source plates. In the latter case, robotic printers containing several print-tips are used, which partition the microarray into discrete subarray blocks representing the different tips. Subsequently, one or two labeled nucleic acid samples are hybridized to the microarray under optimally calibrated conditions and the slide is then scanned to quantify probe-specific intensity calls. The raw data obtained is usually subjected to several steps of quality control and normalization in order to remove possible biases originating in any of above steps ([1–4]).
The reliability of microarray results has been questioned due to inconsistencies in the reported data and in conclusions reached within and between different studies [5–13]. Other studies claim for adequate microarray data reproducibility [14–18]. Recently, the MicroArray Quality Control (MAQC) consortium addressed the reliability of data obtained using microarrays, by directly comparing performance across multiple platforms, test sites and replicates . Concordance of qualitative gene detection calls were around 80–95% for intrasite replicates, 70–85% for intersite replicates, and 60–80% for different platforms. Alternative technologies for quantitative gene expression, such as RT-PCR, seem to provide more reliable results [19, 20]. In addition, many microarray studies do not match the MAQC platform quality, experimentation expertise and relative high signal-to-noise ratios of the samples compared, and would thus generate data of yet poorer reliability. The specific technical sources underlying the suboptimal quality of the microarray technology are unclear; their identification could have a significant impact on genomic research.
Here, we investigated a specific technical effect previously reported to influence microarray data. In certain microarrays, gene probes are printed on the microarray surface according to their chromosomal position or a simple transformation thereof. When coupled to spatial biases, i.e. uneven intensity measurements across the microarray surface, such non-random probe placement designs give rise to spurious correlations between genes at particular relative positions in the genome [21–23]. This was suggested as a possible factor in the reported co-expression of adjacent genes in yeast, originally discovered in a study of gene expression during the cell cycle [24, 25]. It has been suggested that print-tip effects comprise a dominant source of spatial bias underling spurious periodicities in this case . Consistently, common normalization practices correct for print-tip effects () or ignore spatial biases altogether. Another study showed that inadequate cleaning of print-tips causes "carry-over" during the printing process and contributes to the generation of spurious correlations between adjacent probes . However, print-tip-related effects are irrelevant to in-situ printed microarrays, which nonetheless exhibit spurious chromosomal-position-dependent correlations. This indicates that additional or different sources of bias are responsible for spurious correlations observed in gene expression studies.
In order to assess the extent of the effect causing spurious correlations in yeast microarray studies, we applied an autocorrelation analysis on a database of over 2000 individual microarray experiments. Remarkably, we find that spurious periodicities dominate yeast microarray datasets. Moreover, we demonstrate that they result from large and continuous spatial biases on the microarray surface, which are generated at the microarray hybridization step. The extent of such spatial biases, which are probably ubiquitous in microarray studies, has not been previously appreciated. We also show that autocorrelation can be used for the identification of aneuploidies in the strains used for expression studies, and that in certain cases, conclusions regarding segmental genetic changes can also arise spuriously.
Results and discussion
Autocorrelation analysis reveals spurious periodicities dependent on microarray design
We tested the utility of the autocorrelation analysis on two cell cycle experiments [24, 27] reported to exhibit the microarray design-related effect [21, 22]. As expected, we observed very strong autocorrelation signals (Figure 1A). While autocorrelation values for gene distances of up to ~5 were highest, secondary peaks in the autocorrelation profile were also very clear. Consistent with previous observations [21, 22], the secondary peaks appeared with periodicities of 24 and 13 genes for the two different experiments.
Presence of periodic autocorrelation patterns in multiple experiment sets
The fact that autocorrelation periodicities appear in diverse datasets prompted us to assess their extent over a wide range of microarray studies. We assembled a set of 2005 yeast microarray experiments from different laboratories, platforms and experimental procedures (see materials and methods). Numerous periodic autocorrelation patterns were observed in these experiments (Figure 3F). We quantified the extent of these periodicities by performing a second iteration of autocorrelation, which greatly enhances any periodic signals while having a minor effect on other signals, and by defining a strict significance criterion of over 20 signal points with autocorrelation r values greater than 0.05 at gene distances of up to 200. We found that 1194 of 2005 (59.5%) experiments passed this significance criterion, which is associated with a P value smaller than 10-16. We consider this percentage a lower bound for the fraction of experiments suffering from periodic autocorrelations. Only one source of microarrays, those produced by Rosetta Inpharmatics, did not display such periodicities, presumably due to a random probe placement design (Figure 3G). None of the 340 microarrays from this set passed our significance test for periodic autocorrelation. We conclude that the cause of spurious autocorrelations observed in the cell cycle studies dominates yeast microarray studies, and that this bias influences the final data to an extent that it can be observed as significant autocorrelation periodicities. Such spurious correlations are not confined to yeast microarrays, as they were also reported to occur in C. elegan s and human microarray experiments .
Widespread spatial biases in microarray experiments
The argument that large, print-tip independent spatial biases are the cause of spurious periodicities is also consistent with the stochastic, rather than systematic nature of the appearance of periodicities (Figure 3). It is also supported by the presence of spurious periodicities in Affymetrix microarray experiments (Figure 3B), which do not contain subarrays and in which no source plates or printing tips are used. These attributes are consistent with random hybridization inhomogeneities serving as the source of spatial biases. We note that the SMD yeast microarray design is composed of only four subarray blocks, which complicated the distinction between a subarray effect and other biases and led to the previous attribution of print-tip effects to spurious correlations . Accordingly, print-tip normalization seems inappropriate for correction of spatial biases, and may instead introduce unwanted edge-effects. An additional contributing factor to spatial biases was suggested to be a "carry-over" caused by inappropriate cleaning of print-tips between probe printings . However, this bias produces only 0.1% noise for fully-hybridized probe spots and is at most a negligible factor relative to large spatial biases. Our simulations, which were performed on a background of random data, demonstrate that large and continuous spatial biases could solely explain all of the observed spurious periodicities.
The occurrence of large spatial biases in microarray experiments from both yeast and other organisms was previously reported ([1, 28–30]). However, the use of autocorrelation analysis on data obtained from microarrays printed in a non-random manner with respect to chromosomal position has enabled us to quantify the extent of such biases over multiple experiments. We accordingly demonstrate that spatial biases occur in a majority of microarray experiments. The prevalence of such biases is probably even higher than estimated by our autocorrelation analyses, which do not detect weak or small-sized spatial biases. Importantly, the same extent of spatial biases could be expected to occur regardless of microarray design, although autocorrelation would not be useful for their identification in such cases. Since we analyzed experiments from a variety of platforms, laboratories and procedures, we infer that spatial biases are a ubiquitous characteristic of microarray studies in general.
The above conclusions emphasize the need to apply a spatial bias correction step when analyzing microarray data. We tested several methods for spatial bias correction and found that virtually any method, including print-tip normalization and corrections of spatial gradients, effectively eliminate all periodic autocorrelation signals (data not shown). However, none of these capture the actual nature of the spatial trends and can introduce additional biases and edge effects. Instead, a method termed MANOR (Micro-Array NORmalization) has previously been presented , which accounts for both local, abrupt spatial signal changes, as well as continuous intensity gradients. MANOR combines a spatial segmentation procedure with a two-dimensional Loess regression and is optimized to preserve the true biological signal when correcting for spatial biases. It is publicly implemented in an R package (available at ). We consider MANOR the most suitable algorithm for the correction of spatial biases in microarray experiments in general. Although originally implemented in spatial normalization of array-CGH data, our demonstration of widespread spatial biases in various sorts of microarray experimental procedures makes it relevant also to non-CGH experiments, for which it comparably removes autocorrelation periodicities (not shown).
Identification of aneuploidy by autocorrelation
Several of the long autocorrelation tracts we observed in our expression data assembly may represent additional cases of aneuploidies in the strains used for generating the data. However, others occur in experiments in which the control and experiment samples were taken from genetically-identical culture samples. We suspected that the autocorrelation patterns observed in these cases may be the result of another ramification of the effect of specific spatial biases coupled to microarray design. Indeed, we found that narrow and long spatial biases in the SMD microarray design can cause such an aneuploidy-like signal, which is nonetheless spurious in origin (Figure 7C). Thus, spatial biases can lead to false identification of genetic alterations in studies based on non-random microarray designs.
In this study we have demonstrated the utility of autocorrelation analysis for the efficient identification and filtering of spurious chromosomal-position-dependent correlations. In particular, we provide compelling evidence for the prevalence of large spatial biases in microarray studies, to an extent unappreciated thus far. Our conclusions are based on data simulations, the stochastic nature of spurious autocorrelation patterns, and the existence of spurious correlations in spotted as well as Affymetrix microarray experiments. Although we have identified spatial biases by their manifestation in the form of periodic autocorrelation, which in itself depends on microarray design, their frequency of occurrence should be constant over many microarray platforms irrespective of design. Our simulations suggest that spatial biases are commonly associated with signal changes of a factor of two or more over large portions of the data, which represents a significant extent of bias and a potent source of false data. Spatial biases can accordingly explain the many cases of poor or suboptimal reproducibility in microarray studies. We suggest that normalization methods that correct for spatial biases, such as MANOR , should be routinely applied when analyzing microarray data. Re-analysis of existing data should also consider such spatial biases and their effect on the data. Finally, future improvement of microarray data quality should concentrate on overcoming spatial biases, mainly by optimization of hybridization procedures.
BY4743 diploid Saccharomyces cerevisiae cells grown in YPD media were arrested in late G2 by addition of 10 μg/ml Nocodazole (Sigma) for 1.5 hours and subsequently released into the cell cycle. Sample preparation, microarray hybridization and data extraction were performed as previously described . The data was background subtracted and not normalized for print-tip-dependent or other spatial biases. Microarrays used were the UHN Y6.4k4 PCR-product microarrays representing complete yeast ORFs (University Health Network, Toronto), and the UMC Utrecht S. cerevisiae 16K array version 1.1, which consist of 70-mer oligonucleotide probes unique for each yeast gene. The raw data and log2-transformed ratio data ordered by genomic position, for each of the microarray designs, can be found at our website, at .
External datasets used
All public data analyzed is background-subtracted intensity or ratio calls without any spatial bias normalization.
Yeast cell cycle expression data corresponds to the α-factor arrest and release experiment from Spellman et al., 1998 , hybridized onto SMD Saccharomyces cerevisiae Array y744, and the cdc15 temperature-sensitive mutant arrest and release experiments from Cho et al., 1998 , hybridized onto Affymetrix YE6100 microarrays. For Figure 1, we used the 42 and 110 minute time points from these studies, respectively.
We analyzed a total of 2438 separate microarray experiments from the following sources: 1) A previously described yeast gene expression database (, details of which can be found at ), which was assembled in 2002 and includes experiments performed on a variety of microarray platforms, including 125 experiments from early versions (YE6100 and S98) of Affymetrix yeast expression microarrays. 2) The complete database of yeast microarray studies from the Stanford Microarray Database (SMD; [38, 39]) excluding experiments already included in the former database and those with >1000 missing values. This database also covers experiments recently performed. 3) A set of 113 ChIP-on-chip experiments . We used the P value data for this experimental set since it constitutes the relevant user-level data for these experiments; the ratio data from which the P value data was derived yielded the same results in terms of autocorrelations. These experiments add to another 83 ChIP-on-chip experiments from the SMD database.
Experiments performed with deletion strains harboring verified aneuploidies , as well as the ymr031w-a deletion strain, and comparative genome hybridization (CGH) experiments from the SMD database gave a unique autocorrelation signature and were analyzed separately. An additional 340 experiments from three studies performed on microarrays designed by Rosetta Inpharmatics [33, 41, 42] showed no autocorrelation patterns, presumably due to random probe placement (we could not verify this), and were thus separated from the rest of the database and treated as a negative control.
The log2-transformed ratio data (or intensity in Affymetrix experiments) was used for autocorrelation analyses. Genes were ordered according to their genomic position, taken from the Saccharomyces Genome Database (SGD; ). Pearson correlation coefficients were determined for distances of between one gene and the size of the gene list-1, according to the formula: Autocorr(X, i) = Corr(X(1:L-i), X(i:L)), where X is the ordered data, i is the gene distance, and L is the length of the gene list. Missing values in the data were given a log2 ratio value of zero; this caused a decrease in the autocorrelations values to some extent, but retained the actual periods themselves.
In order to evaluate the significance of periodic patterns in the autocorrelations, we performed an autocorrelation analysis on the autocorrelation data itself. Any periodic signals are significantly enhanced by this procedure, while having only a marginal effect on non-periodic signals. We chose a significance criterion of second-iteration autocorrelation r values of >0.05, and demanded that at least 20 data points out of the first 200 pass this criterion in order for an experiment to be regarded as containing significant periodicities. These figures were chosen since they yielded zero false positives in the control dataset (Figure 3G) and, by visual inspection, identified the maximal number of true periodicities in the studied dataset. The P value of this criterion is <10-16 (using the binomial distribution on randomized autocorrelation data, which distributed approximately normally with mean ~0 and standard deviation ~0.01).
Simulations of spatial biases
Random expression data was generated by permutating measurements from a given experiment. Either individual subarray blocks, or circular spatial shapes, were given a ten-fold higher value in one channel. Circular shapes were defined as complying to the formula: , where X and Y are the coordinates of the spots that fall within the bias shape, C1 and C2 represent the center coordinates (set at the center of the microarray surface), F is a circularity factor (set at 1 for horizontally-shaped circular biases and for generating Figure 5, and at 8 for vertically-shaped biases), and R is the radius of the bias. Subsequently, autocorrelations were calculated on the log2-transformed ratio data for each introduced bias.
We are grateful to Frank Holstege and the UMC Utrecht genomics laboratory personnel for microarrays and generous assistance. We thank Adina Weinberger for participation in the cell cycle experiments, Ilya Soifer for help with data analysis, and Judith Berman and members of our laboratory for helpful discussions. This work was supported by grants from the Tauber fund, the Kahn fund for Systems Biology at the Weizmann Institute of Science, and the Israeli Ministry of Science (Tashtiot program).
- Eads B, Cash A, Bogart K, Costello J, Andrews J: Troubleshooting Microarray Hybridizations. Methods in Enzymology. 2006, 411: 34-49.PubMedView ArticleGoogle Scholar
- Quackenbush J: Microarray data normalization and transformation. Nat Genet. 2002, 32 Suppl: 496-501. 10.1038/ng1032.PubMedView ArticleGoogle Scholar
- Tseng GC, Oh MK, Rohlin L, Liao JC, Wong WH: Issues in cDNA microarray analysis: quality filtering, channel normalization, models of variations and assessment of gene effects. Nucl Acids Res. 2001, 29: 2549-2557. 10.1093/nar/29.12.2549.PubMed CentralPubMedView ArticleGoogle Scholar
- Yang YH, Dudoit S, Luu P, Lin DM, Peng V, Ngai J, Speed TP: Normalization for cDNA microarray data: a robust composite method addressing single and multiple slide systematic variation. Nucl Acids Res. 2002, 30: e15-10.1093/nar/30.4.e15.PubMed CentralPubMedView ArticleGoogle Scholar
- Grunenfelder B, Winzeler EA: Treasures and traps in genome-wide data sets: case examples from yeast. Nat Rev Genet. 2002, 3: 653-661. 10.1038/nrg886.PubMedView ArticleGoogle Scholar
- Tan PK, Downey TJ, Spitznagel EL, Xu P, Fu D, Dimitrov DS, Lempicki RA, Raaka BM, Cam MC: Evaluation of gene expression measurements from commercial microarray platforms. Nucl Acids Res. 2003, 31: 5676-5684. 10.1093/nar/gkg763.PubMed CentralPubMedView ArticleGoogle Scholar
- Miklos GLG, Maleszka R: Microarray reality checks in the context of a complex disease. Nat Biotech. 2004, 22: 615-621. 10.1038/nbt965.View ArticleGoogle Scholar
- Marshall E: Getting the Noise Out of Gene Arrays. Science. 2004, 306: 630-631. 10.1126/science.306.5696.630.PubMedView ArticleGoogle Scholar
- Steinmetz LM, Davis RW: Maximizing the potential of functional genomics. Nat Rev Genet. 2004, 5: 190-201. 10.1038/nrg1293.PubMedView ArticleGoogle Scholar
- Frantz S: An array of problems. Nat Rev Drug Discov. 2005, 4: 362-363. 10.1038/nrd1746.PubMedView ArticleGoogle Scholar
- Michiels S, Koscielny S, Hill C: Prediction of cancer outcome with microarrays: a multiple random validation strategy. The Lancet. 2005, 365: 488-492. 10.1016/S0140-6736(05)17866-0.View ArticleGoogle Scholar
- Tibshirani R, Hong WJ, Warnke R, Chu G, Staudt LM, Wright G, Dave S: Immune Signatures in Follicular Lymphoma. N Engl J Med. 2005, 352: 1496-1497. 10.1056/NEJM200504073521422.PubMedView ArticleGoogle Scholar
- Ein-Dor L, Zuk O, Domany E: Thousands of samples are needed to generate a robust gene list for predicting outcome in cancer. PNAS. 2006, 103: 5923-5928. 10.1073/pnas.0601231103.PubMed CentralPubMedView ArticleGoogle Scholar
- Petersen D, Chandramouli GV, Geoghegan J, Hilburn J, Paarlberg J, Kim CH, Munroe D, Gangi L, Han J, Puri R, Staudt L, Weinstein J, Barrett JC, Green J, Kawasaki ES: Three microarray platforms: an analysis of their concordance in profiling gene expression. BMC Genomics. 2005, 6: 63-10.1186/1471-2164-6-63.PubMed CentralPubMedView ArticleGoogle Scholar
- Dobbin KK, Beer DG, Meyerson M, Yeatman TJ, Gerald WL, Jacobson JW, Conley B, Buetow KH, Heiskanen M, Simon RM, Minna JD, Girard L, Misek DE, Taylor JMG, Hanash S, Naoki K, Hayes DN, Ladd-Acosta C, Enkemann SA, Viale A, Giordano TJ: Interlaboratory Comparability Study of Cancer Gene Expression Analysis Using Oligonucleotide Microarrays. Clin Cancer Res. 2005, 11: 565-572.PubMedGoogle Scholar
- Irizarry RA, Warren D, Spencer F, Kim IF, Biswal S, Frank BC, Gabrielson E, Garcia JGN, Geoghegan J, Germino G, Griffin C, Hilmer SC, Hoffman E, Jedlicka AE, Kawasaki E, Martinez-Murillo F, Morsberger L, Lee H, Petersen D, Quackenbush J, Scott A, Wilson M, Yang Y, Ye SQ, Yu W: Multiple-laboratory comparison of microarray platforms. Nat Methods. 2005, 2: 345-350. 10.1038/nmeth756.PubMedView ArticleGoogle Scholar
- Larkin JE, Frank BC, Gavras H, Sultana R, Quackenbush J: Independence and reproducibility across microarray platforms. Nat Methods. 2005, 2: 337-344. 10.1038/nmeth757.PubMedView ArticleGoogle Scholar
- Kuo WP, Liu F, Trimarchi J, Punzo C, Lombardi M, Sarang J, Whipple ME, Maysuria M, Serikawa K, Lee SY, McCrann D, Kang J, Shearstone JR, Burke J, Park DJ, Wang X, Rector TL, Ricciardi-Castagnoli P, Perrin S, Choi S, Bumgarner R, Kim JH, Short GF, Freeman MW, Seed B, Jensen R, Church GM, Hovig E, Cepko CL, Park P, Ohno-Machado L, Jenssen TK: A sequence-oriented comparison of gene expression measurements across different hybridization-based technologies. Nat Biotechnol. 2006, 24: 832-840. 10.1038/nbt1217.PubMedView ArticleGoogle Scholar
- Shi L, Reid LH, Jones WD, MAQCconsortium: The MicroArray Quality Control (MAQC) project shows inter- and intraplatform reproducibility of gene expression measurements. Nat Biotechnol. 2006, 24: 1151-1161. 10.1038/nbt1239.PubMedView ArticleGoogle Scholar
- Canales RD, Luo Y, Willey JC, Austermiller B, Barbacioru CC, Boysen C, Hunkapiller K, Jensen RV, Knight CR, Lee KY, Ma Y, Maqsodi B, Papallo A, Peters EH, Poulter K, Ruppel PL, Samaha RR, Shi L, Yang W, Zhang L, Goodsaid FM: Evaluation of DNA microarray results with quantitative gene expression platforms. Nat Biotech. 2006, 24: 1115-1122. 10.1038/nbt1236.View ArticleGoogle Scholar
- Kluger Y, Yu H, Qian J, Gerstein M: Relationship between gene co-expression and probe localization on microarray slides. BMC Genomics. 2003, 4: 49-10.1186/1471-2164-4-49.PubMed CentralPubMedView ArticleGoogle Scholar
- Balazsi G, Kay KA, Barabasi AL, Oltvai ZN: Spurious spatial periodicity of co-expression in microarray data due to printing design. Nucl Acids Res. 2003, 31: 4425-4433. 10.1093/nar/gkg485.PubMed CentralPubMedView ArticleGoogle Scholar
- Yu H, Nguyen K, Royce T, Qian J, Nelson K, Snyder M, Gerstein M: Positional artifacts in microarrays: experimental verification and construction of COP, an automated detection tool. Nucl Acids Res. 2006, gkl871-Google Scholar
- Cho RJ, Campbell MJ, Winzeler EA, Steinmetz L, Conway A, Wodicka L, Wolfsberg TG, Gabrielian AE, Landsman D, Lockhart DJ, Davis RW: A Genome-Wide Transcriptional Analysis of the Mitotic Cell Cycle. Molecular Cell. 1998, 2: 65-73. 10.1016/S1097-2765(00)80114-8.PubMedView ArticleGoogle Scholar
- Cohen BA, Mitra RD, Hughes JD, Church GM: A computational analysis of whole-genome expression data reveals chromosomal domains of gene expression. Nat Genet. 2000, 26: 183-186. 10.1038/79896.PubMedView ArticleGoogle Scholar
- Gottman JM: Time-series analysis: A comprehensive introduction for social scientists. New York: Cambridge University Press;. 1981Google Scholar
- Spellman PT, Sherlock G, Zhang MQ, Iyer VR, Anders K, Eisen MB, Brown PO, Botstein D, Futcher B: Comprehensive Identification of Cell Cycle-regulated Genes of the Yeast Saccharomyces cerevisiae by Microarray Hybridization. Mol Biol Cell. 1998, 9: 3273-3297.PubMed CentralPubMedView ArticleGoogle Scholar
- Futschik ME, Crompton T: OLIN: optimized normalization, visualization and quality testing of two-channel microarray data. Bioinformatics. 2005, 21: 1724-1726. 10.1093/bioinformatics/bti199.PubMedView ArticleGoogle Scholar
- Neuvial P, Hupe P, Brito I, Liva S, Manie E, Brennetot C, Radvanyi F, Aurias A, Barillot E: Spatial normalization of array-CGH data. BMC Bioinformatics. 2006, 7: 264-10.1186/1471-2105-7-264.PubMed CentralPubMedView ArticleGoogle Scholar
- Reimers M, Weinstein JN: Quality assessment of microarrays: visualization of spatial artifacts and quantitation of regional biases. BMC Bioinformatics. 2005, 6: 166-10.1186/1471-2105-6-166.PubMed CentralPubMedView ArticleGoogle Scholar
- Bioconductor. [http://www.bioconductor.org]
- Hughes TR, Roberts CJ, Dai H, Jones AR, Meyer MR, Slade D, Burchard J, Dow S, Ward TR, Kidd MJ, Friend SH, Marton MJ: Widespread aneuploidy revealed by DNA microarray expression profiling. Nat Genet. 2000, 25: 333-337. 10.1038/77116.PubMedView ArticleGoogle Scholar
- Hughes TR, Marton MJ, Jones AR, Roberts CJ, Stoughton R, Armour CD, Bennett HA, Coffey E, Dai H, He YD: Functional Discovery via a Compendium of Expression Profiles. Cell. 2000, 102: 109-126. 10.1016/S0092-8674(00)00015-5.PubMedView ArticleGoogle Scholar
- Tirosh I, Weinberger A, Carmi M, Barkai N: A genetic signature of interspecies variations in gene expression. Nat Genet. 2006, 38: 830-834. 10.1038/ng1819.PubMedView ArticleGoogle Scholar
- Naama Barkai lab Autocorrelations. [http://barkai-serv.weizmann.ac.il/autocorrelations/]
- Ihmels J, Friedlander G, Bergmann S, Sarig O, Ziv Y, Barkai N: Revealing modular organization in the yeast transcriptional network. Nat Genet. 2002, 31: 370-377.PubMedGoogle Scholar
- Naama Barkai lab Modules. [http://www.weizmann.ac.il/home/jan/NG/MainFrames.html]
- Stanford Microarray Database. [http://genome-www5.stanford.edu/]
- Sherlock G, Hernandez-Boussard T, Kasarskis A, Binkley G, Matese JC, Dwight SS, Kaloper M, Weng S, Jin H, Ball CA, Eisen MB, Spellman PT, Brown PO, Botstein D, Cherry JM: The Stanford Microarray Database. Nucl Acids Res. 2001, 29: 152-155. 10.1093/nar/29.1.152.PubMed CentralPubMedView ArticleGoogle Scholar
- Lee TI, Rinaldi NJ, Robert F, Odom DT, Bar-Joseph Z, Gerber GK, Hannett NM, Harbison CT, Thompson CM, Simon I, Zeitlinger J, Jennings EG, Murray HL, Gordon DB, Ren B, Wyrick JJ, Tagne JB, Volkert TL, Fraenkel E, Gifford DK, Young RA: Transcriptional Regulatory Networks in Saccharomyces cerevisiae. Science. 2002, 298: 799-804. 10.1126/science.1075090.PubMedView ArticleGoogle Scholar
- Marton MJ, DeRisi JL, Bennett HA, Iyer VR, Meyer MR, Roberts CJ, Stoughton R, Burchard J, Slade D, Dai H, Bassett DE, Hartwell LH, Brown PO, Friend SH: Drug target validation and identification of secondary drug target effects using DNA microarrays. Nat Med. 1998, 4: 1293-1301. 10.1038/3282.PubMedView ArticleGoogle Scholar
- Roberts CJ, Nelson B, Marton MJ, Stoughton R, Meyer MR, Bennett HA, He YD, Dai H, Walker WL, Hughes TR, Tyers M, Boone C, Friend SH: Signaling and Circuitry of Multiple MAPK Pathways Revealed by a Matrix of Global Gene Expression Profiles. Science. 2000, 287: 873-880. 10.1126/science.287.5454.873.PubMedView ArticleGoogle Scholar
- Saccharomyces Genome Database. [http://www.yeastgenome.org/]
- Hardwick JS, Kuruvilla FG, Tong JK, Shamji AF, Schreiber SL: Rapamycin-modulated transcription defines the subset of nutrient-sensitive signaling pathways directly controlled by the Tor proteins. PNAS. 1999, 96: 14866-14870. 10.1073/pnas.96.26.14866.PubMed CentralPubMedView ArticleGoogle Scholar
- Posas F, Chambers JR, Heyman JA, Hoeffler JP, de Nadal E, Arino J: The Transcriptional Response of Yeast to Saline Stress. J Biol Chem. 2000, 275: 17249-17255. 10.1074/jbc.M910016199.PubMedView ArticleGoogle Scholar
- Dunham MJ, Badrane H, Ferea T, Adams J, Brown PO, Rosenzweig F, Botstein D: Characteristic genome rearrangements in experimental evolution of Saccharomycescerevisiae. PNAS. 2002, 99: 16144-16149. 10.1073/pnas.242624799.PubMed CentralPubMedView ArticleGoogle Scholar
- Dunn B, Levine RP, Sherlock G: Microarray karyotyping of commercial wine yeast strains reveals shared, as well as unique, genomic signatures. BMC Genomics. 2005, 6: 53-10.1186/1471-2164-6-53.PubMed CentralPubMedView 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.