PLoS One. 2012; 7(4): e35147

MicroRNA screening RNA was extracted from AGS cells stably expressing CagA or the empty vector using Trizol reagent (Invitrogen) and was sent to CapitalBio, Beijing, for mammalian miRNA expression profile microarray detection, which included 1024 miRNAs from mouse, rat, and human. The results were compared with those from the control group; changes that were >2-fold or <2-fold were considered significant.

PLoS Med. 2006 October; 3(10): e420.

siRNA Duplex Screening and TransfectionThe siRNA sequence targeting NRF2 corresponds to the coding region nucleotides 1903–1921 (5′-GTAAGAAGCCAGATGTTAA-3′) in the NRF2 cDNA. The NRF2 siRNA duplex with the following sense and antisense sequences was used: 5′-GUAAGAAGCCAGAUGUUAAdUdU-3′ (sense) and 3′-dUdUCAUUCUUCGGUCUACAATT-5′ (antisense). KEAP1 siRNA corresponds to the coding region nucleotides 1545–1563 (5′-GGGCGTGGCTGTCCTCAAT-3′) in KEAP1 transcript variant 2. The KEAP1 siRNA duplex with the following sense and antisense sequences was used: 5′-GGGCGUGGCUGUCCUCAAUdUdU-3′ (sense) and 3′-dUdUCCCGCACCGACAGGAGUUA-5′ (antisense). To confirm the specificity of the inhibition, the siCONTROL non-targeting siRNA 1 (NS siRNA; 5′-UAGCGACUAAACACAUCAAUU-3′) with microarray-defined signature was used as a negative control. All of the siRNA duplexes were synthesized by Dharmacon Research (Lafayette, Colorado, United States). Cells in the exponential growth phase were plated at a density of 0.2 × 106 cells/ml, grown for 12 h, and transfected twice at an interval of 48 h with 50 nM siRNA duplexes using Lipofectamine 2000 and OPTI-MEM reduced serum medium (Invitrogen) according to the manufacturer's recommendations. Concentrations of siRNAs were chosen on the basis of dose–response studies (data not shown).

PLoS One. 2012; 7(6): e38716.

miRNA MicroarrayTotal RNAs of the two littermate piglets were pooled for the microarray analysis. The pooled samples were named as EHL1–3 (n=3) and LW1-3 (n=3).The pig microRNA microarray was obtained from LC Sciences (Houston, USA) and contains 238 unique probes that were complementary to all mature miRNAs of pig in miRBase release 16.0. The assay started from 5 µg total RNA sample, which was size fractionated using a YM-100 Microcon centrifugal filter (Millipore, USA) and the small RNAs (<300 nt) isolated were 3′-extended with a poly(A) tail using poly(A) polymerase. An oligonucleotide tag was then ligated to the poly(A) tail for later fluorescent dye staining. Hybridization was performed overnight on a µParaflo microfluidic chip using a micro-circulation pump (Atactic Technologies, USA). On the microfluidic chip, each detection probe consisted of a chemically modified nucleotide coding segment complementary to target miRNAs (from miRBase, release 16, 238 probe sets) and a spacer segment of polyethylene glycol to extend the coding segment away from the substrate. The detection probes were made by in situ synthesis using photogenerated reagent (PGR) chemistry. The hybridization melting temperatures were balanced by chemical modifications of the detection probes. 100 µL 6×SSPE buffer containing 25% formamide was used for hybridization at 34°C. After hybridization, fluorescence labeling using tag-specific Cy5 dyes was used for detection. Hybridization images were collected using a laser scanner (GenePix 4000B, Molecular Device) and digitized using Array-Pro image analysis software (Media Cybernetics). Data were analyzed by first subtracting the background and then normalizing the signals using a LOWESS (locally weighted regression) function. The three samples of LW piglets were used as control group to analysis the different expression between the two pig lines. The differences between the two groups were analyzed using SAM (Significance Analysis of Microarrays) method with SAMR software version 3.02 [42]. The FDR (False Discovery Rate) and fold change were calculated. miRNAs with FDR <15% were considered to be differentially expressed miRNAs. All the microarray data were MIAME compliant and have been deposited in GEO (accession number GSE33523). Real-time qPCR Confirmation of Different Expressed miRNAsTwo µg of total RNA from each piglet were polyadenylated by poly(A) polymerase (PAP) at 37°C for 1 h in a 20 µL reaction mixture using Poly(A) Tailing Kit (AM1350, Ambion, USA) according to the manufacturer’s instructions. Treated RNA was then dissolved and reverse-transcribed using poly (T) adapter and M-MLV (Promega, USA). qPCR for the 18 miRNAs was performed using SYBR Green Real-time PCR Master Mix (TaKaRa, Japan) in Mx3000P (Stratagene, USA) with a miRNA specific forward primer and a universal reverse primer complementary to part of the poly(T) adapter sequence. U6 was chosen as an internal control to normalize the technical variations. The sequences for all the primers and the poly (T) adapter are listed in Table S4. The method of 2−ΔΔCT was used to analyze the real-time PCR data expressed as the fold change relative to the average value of the LW piglets [43]. The differences between the two groups were analyzed by the student t-test for independent samples with the SPSS 13.0 for Windows.

Cancer Cell. 2011 October 18; 20(4): 457–471.

Gene expression microarray assayVCaP or VCS2 cells treated with ethanol or 10nM DHT were subjected to microarray assay (Affymetrix) to identify genes whose expression was repressed by DHT in both VCaP and VCS2 cells. Tissue mRNA was extracted and purified from three sets (pre-castrated, 4d-post-castrated, and relapsed) of xenograft tumors (3 mice) and then subjected to microarray assay (Agilent). SAM software was used to perform t-test on these three biological repeats (three mice) to determine the score and q-value. The genes whose expression was significantly elevated in relapsed tumors (q<0.05) were picked for the next screening to determine if they were DHT-repressed in VCaP and VCS2.

PLoS One. 2012; 7(9): e44737.

Microarray ExperimentsCells samples including hUC-MSCs and hUC-MSCs after hepatic differentiation for 2 days, 6 days, 10 days, 14 days, 22 days and 26 days were collected for microRNA expression analysis with Agilent human miRNA (8*60K) V16.0. Microarray experiments including RNA extraction and purification, RNA labeling, array hybridization and data acquisition were performed at Shanghai Biochip Company according to the protocols for the Agilent miRNA microarray system. Total RNA was extracted and purified with a mirVana™ miRNA Isolation Kit (Ambion, Austin, TX, US) and checked for a RIN number to inspect RNA integration with an Agilent Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA, US). miRNA molecules in the total RNA were labeled with a miRNA Complete Labeling and Hyb Kit (Agilent technologies). Each slide was hybridized with 100 ng Cy3-labeled RNA with a miRNA Complete Labeling and Hyb Kit (Agilent Technologies) in a hybridization oven (Agilent Technologies) at 55°C, and 20 rpm for 20 hours. After hybridization, the slides were washed in staining dishes (Thermo Shandon, Waltham, MA, US) with a Gene Expression Wash Buffer Kit (Agilent Technologies).

Int J Biol Sci. 2012; 8(10): 1363–1374.

Patients and Tissue SamplesA total 22 non-malignant tissue samples from 22 non-metastatic ccRCC patients (21 male, one female; median age 62, range 41-78 years) and 22 malignant samples from another 22 ccRCC patients (21 male, one female; median age 67, range 39-87 years) undergoing radical nephrectomy as well as metastatic tissue samples from 13 patients (12 male, one female, median age 67, range 40-89 years) undergoing surgical resection of ccRCC bone metastases were collected between 2003 and 2010 at the University Hospital Charité, as previously reported 24. Tissue samples were snap frozen in liquid nitrogen directly after surgery and were stored at -80°C until RNA isolation. Fifteen tumor samples were classified as pT1, one as pT2, and six as pT3, according to the 2002 TNM classification, while the histological grading according to the 2004 WHO criteria resulted in four G1, 17 G2, and one G3 (Table ​(Table1).1). All of the tumor types were ccRCC. The study was approved by the Ethics Committee of the University Hospital Charité in accordance with the principles of Declaration of Helsinki; the patients provided informed consent. All of these samples were used for the validation of the microarray-based selection of differentially expressed miRNAs with RT-qPCR. Tissue samples used for microarray profiling were previously described 24.Table 1Clinical and histopathological characteristics of investigated patients.

Low ERK Phosphorylation in Cancer-Associated Fibroblasts Is Associated with Tamoxifen Resistance in Pre-Menopausal Breast Cancer. PLoS ONE 7(9): e45669. doi:10.1371/journal.pone.0045669

Patients and Tumor Samples Breast cancer cohort I includes 564 pre-menopausal patients, enrolled in a trial from 1986 to 1991 and randomized to either 2 years of adjuvant tamoxifen treatment (n = 276) or no systemic treatment (n = 288). All patients were followed up for recurrence-free survival. Recurrence was defined as local, regional, or distant recurrence and breast cancer-specific death, whereas contralateral breast cancer was excluded. Each patient underwent surgery (either modified radical mastectomy or breast conserving surgery) followed by radiotherapy and in a small number of cases adjuvant polychemotherapy (less than 2%). The median post-surgery follow-up time without a breast cancer event was 13.9 years. Further details of the trial have been previously described [13], [14]. Breast cancer cohort II includes 179 pre- and post-menopausal patients undergoing endocrine or chemotherapy, diagnosed with primary invasive breast cancer between 2000 and 2002, at the Department of Pathology, Malmö University Hospital. This cohort was designed as a first-line screening cohort for Human Protein Atlas (HPA) antibodies with potential relevance in breast cancer [15]. Median age at diagnosis was 65 years (range 35–97) and median follow-up time 69 months. All patients in this cohort had received treatment following surgery. For detailed description of clinico-pathological features of the tumor samples we refer to previous studies [16], [17]. Representative tumor areas of formalin-fixed and paraffin-embedded tissue material were selected for tissue microarray (TMA) construction. Details regarding TMA assembling and staining procedure have been reported [13].

Cancer Cell Int. 2007; 7: 14.

In silico analysis of CpG-island association, gene annotation, and pathway enrichment The criteria for a CpG-island was based on those outlined by Takai and Jones [61], where the GC ≥ 55%, Obs/Exp ≥ 0.65, and length > 300 bp which was reported to exclude most Alu-repetitive elements. We identified the genes that harbored CpG-island within a 2000 bp window upstream or downstream from the transcription start site based Human Genome Browser data base [59]. To be certain that there were no CpG island closer to the TSS and gene promoter regions, we submitted the sequences of interest (including a 2000 bp window upstream and downstream from TSS) to the CpG search engine available in reference [61] and verified that there was no CpG islands that are closer to TSS for the genes we tested. Up-regulated genes with CpG-island associations were further analyzed through the Microarray Literature-based Annotation tool MILANO [62] to look for evidence of epigenetic modifications in the literature. MILANO is a web-based tool that allows annotation of lists of genes derived from microarray results by user defined terms [62]. Using MILANO we searched for literature associations between our list of genes and the terms 'epigenetics', 'methylation' 'chromatin modification' 'cancer', and 'disease'. To identify the putative functional pathways for each gene list, we used the functional annotation enrichment tool. This tool utilizes the Gene Ontology database and uses GO Terms to identify enriched biological themes in the gene lists [63,63]. The Fisher Exact test was applied to determine the significance in the proportions of genes falling into a certain pathway in each gene list. We used this tool to look for enriched pathways of up- or down- regulated genes with CpG-island associations from the gene lists from the cell lines.

Clin Cancer Res. 2006 June 15; 12(12): 3856–3863

Algorithmic image analysis Images were analyzed using algorithms that have previously been extensively described (26). Two images (one in-focus and one out-of-focus) were taken of the compartment specific tags and the target marker. A percentage of the out-of-focus image was subtracted from the in-focus image for each pixel, representing the signal-to-noise ratio of the image. An algorithm described as Rapid Exponential Subtraction Algorithm was used to subtract the out-of-focus information in a uniform fashion for the entire microarray. Subsequently, the Pixel Locale Assignment for Compartmentalization of Expression algorithm was used to assign each pixel in the image to a specific subcellular compartment and the signal in each location is calculated. Pixels that cannot accurately be assigned to a compartment were discarded. The data were saved and subsequently expressed as the average signal intensity per unit of compartment area. For the nuclear and membrane/cytoplasmic compartments, the image was measured on a scale of 0 to 255 and expressed as target signal intensity relative to the compartment area.

Clin Cancer Res. 2006 June 15; 12(12): 3856–3863

Tissue microarray construction The melanoma tissue microarrays were constructed as previously described (30). A total of 232 primary melanomas, 15 local recurrences, and 299 metastatic cores, each measuring 0.6 mm in diameter, were spaced 0.8 mm apart on glass slides. The cohort was constructed from paraffin-embedded, formalin-fixed tissue blocks obtained from the Yale University Department of Pathology Archives. Specimens and clinical information were collected under the guidelines and approval of a Yale University Institutional Review Board. The cohort has been used in prior publications (32). The specimens were resected between 1959 and 2000, with a follow-up range between 2 months and 40 years (mean follow-up time, 6.7 years). Age at diagnosis ranged from 18 to 91 years (mean age, 52.4 years). The cohort included 55% males and 45% females. The time between tumor resection and tissue fixation was not available. A pathologist reviewed slides from all of the blocks to select representative areas of invasive tumor to be cored. The cores were placed on the tissue microarray using a Tissue Microarrayer (Beecher Instruments, Silver Spring, MD). The tissue microarrays were then cut to 0.5-μm sections and placed on glass slides using an adhesive tape-transfer system (Instrumedics, Inc., Hackensack, NJ) with UV cross-linking. Similarly, a tissue microarray was made containing cores from 540 benign nevi. The nevi array contained 31 metastatic specimens from patients that were also represented on the melanoma array. Both arrays contained identical cell lines, cored from pellets as previously described (33). The overlapping metastatic specimens and cell lines were used for normalization of the scores obtained from the benign and malignant arrays.

J Clin Invest. 2008 January 2; 118(1): 111–123.

Microarray analysis. Total RNA was extracted from frozen tissues of 5 Eμ-myc lymphomas per TRAIL-R genotype (i.e., WT, TRAIL-R+/–, and TRAIL-R–/–) using the RNeasy mini kit (Qiagen). Complementary DNA was synthesized from total RNA using a dT primer tagged with a T7 promoter. Complementary RNA was synthesized by transcription in vitro and labeled with biotinylated nucleotides (Enzo Biochem). All hybridizations were performed using the MOE430A 2.0 GeneChip (Affymetrix). Gene lists, filtered by a minimum of 2-fold change (1-way ANOVA, P < 0.05), were generated with the GeneSpring GX 7.3.1 software (Agilent), which resulted in the creation of gene structures specific to WT and TRAIL-R–deficient mice (total of 59 genes, 17 upregulated and 42 downregulated). Also, more stringent exclusion criteria were applied by the use of significance analysis of microarrays (SAM) software version 2.0 (with an approximate 26.7% false discovery rate) and filtering against gene lists for signal transduction pathways, apoptosis-related genes, and oncogenes obtained from the Gene Ontology Consortium in order to detect 6 differentially expressed genes.

j gen intern med. 2008 january 23(suppl 1): 78–84.

Analytical Tools in Proteomics Method Principle Advantages Disadvantages Electrophoresis Electrophoresis When an electric field is applied to a solution containing a protein that has a net positive or negative charge, the protein migrates at a rate that depends on its net charge, size, and shape. Gels must be stained before proteins can be visualized. Rarely useful by itself as proteins cannot be accurately identified without the use of another detection technique such as immunoblotting or mass spectroscopy. SDS-PAGE Proteins migrate through inert matrix gel of polyacrylamide. Pore size is adjustable to retard protein of interest. SDS is a negatively charged detergent that unfolds proteins and frees them from other molecules. Proteins migrate at different rates toward positive electrode. Separates all types of proteins, even those insoluble in water. One-dimensional separation method has limited resolution. Closely spaced bands or peaks tend to overlap. Can only resolve a small number of proteins. Two-dimensional gel electrophoresis Combines 2 separation procedures. First dimension: the solubilized, denatured proteins are separated by their isoelectric point (pH where net charge is 0) in a polyacrylamide gel. Second dimension: the narrow gel containing proteins separated by isoelectric focusing undergoes electrophoresis at a right angle in SDS-PAGE to separate by size. Good resolution of mixture. Comparison of multiple gels facilitated by image analysis software. Posttranslational modifications can be discerned. Resolution of protein approximately 1ng/mL. Presence of high abundance proteins (i.e., albumin, immunoglobulins) may obscure low abundance proteins. Low throughput. Final identification of protein requires spot removal from gel, digestion, and analysis of peptides by mass spectrometry. Unable to resolve low molecular weight proteins (<10,000Da). Not easily amenable to multivariate analysis. Two-dimensional fluorescence difference gel electrophoresis Labels complex mixtures with fluorescent dyes before conventional two-dimensional electrophoresis. Different cyanine dyes are used to label protein from different samples and will be excited and emit at different light wavelengths. Up to three different samples can be labeled and mixed together (test, control, reference). Analysis of differences between mixtures is simplified. Ratio of protein expression can be obtained in a single gel, and an internal standard can be used in each gel to reduce gel-to-gel variation. Very sensitive. Presence of high abundance proteins (i.e., albumin, immunoglobulins) may obscure low abundance proteins. Low throughput. Final identification of protein requires spot removal from gel, digestion, and analysis of peptides by mass spectrometry. Many spots cannot be identified because of lack of material. Unable to resolve low molecular weight proteins (<10,000Da). Protein array Protein arrays Multiplex protein arrays, cytokine arrays, tissue microarrays In most common form, antibodies to known proteins are tethered to a surface (beads, nitrocellulose, etc.) and then detected using principles of immunoassays. High sensitivity and throughput. Multiple analytes can be measured simultaneously. Identification of potential targets already known. Limited antibody availability and specificity. Required some prior knowledge of expressed proteins. May not detect isoforms of analyte. Cost per sample may be prohibitive. Mass spectroscopy (MS) MS Matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF-MS) Determines the precise mass of protein or peptide fragment from protein. Protein/peptide samples are mixed with organic acid matrix, dried on metal slide, and blasted by laser ionizing the peptide, which is accelerated in an electric field toward a detector. The time it takes to reach the detector is determined by the charge and mass. Peptide sequence information can be obtained with tandem mass spectrometers (MS–MS). Highest resolution is for molecules <3,000Da in size. Surface-enhanced laser desorption/ionization time-of-flight mass spectrometry (SELDI-TOF-MS) Comparable to MALDI, the difference being that SELDI uses chromatographic chip arrays to selectively bind subsets of proteins from complex samples. The surfaces can be washed to remove nonspecifically bound proteins and substances that can interfere with the ionization process (salt, detergents, etc.). High throughput via automation. Requires minimal sample preparation. Can be combined with prefractionation of material to enhance the detection of lower abundant proteins. No direct identification of proteins. Less sensitive to high molecular weight protein (>20kDa). May have instrument-to-instrument variation. Stable isotope labeling Biological samples are labeled with different stable isotopes using modifying agents targeting a specific amino acid (e.g., ICAT). After separation and mass spectrometry, peptides from the 2 samples differing in mass units specific for the isotope used (e.g., 8-Da mass shift for ICAT) can then be used to provide relative quantification. Wider proteome coverage than other methods. Can obtain quantitative information on a large number of proteins; Usually yields IDs of relevant proteins. Technically demanding; very low throughput capability; samples need to be trypsinized before analysis; reliable quantitative measurements likely on most abundant proteins. Adapted from Hoehn and Suffredini.12

plos genet. 2006 november 2(11): e196

Drosophila SL2 cell RNAi. All RNAi for Drosophila SL2 cells was performed as described [52]. Double-stranded RNA was synthesized with T7 RiboMax (Promega, http://www.promega.com). Cells were RNAi depleted using 50 μg of dsRNA for each gene and normalized with luciferase dsRNA for co-RNAi treatments. For RNAi-transfection experiments, 15 μg of dsRNA was included in each transfection after the initial RNAi depletion. Depletion of RBF1 was confirmed by Western analysis (unpublished data). Depletion of Aac11 RNA was confirmed on microarray analysis (unpublished data).

Reproductive Biology and Endocrinology 2008, 6:41

Affymetrix GeneChip technology and bio-informatics RNA was purified using the RNeasy Mini Kit (QIAGEN, Valencia, CA) followed by on-column DNA digestion with RNase-Free DNase (QIAGEN). RNA integrity was confirmed using the RNA 6000 Nano LabChip (Agilent Biotechnologies, Palo Alto, CA). Two replicate cRNA targets were generated from each HEC-1-A sub-line RNA preparation; all eight targets were made at the same time. Total cellular RNA (8 ug) was converted to cDNA using a T7-(deoxythymidine)24 primer and Superscript II (Life Technologies, Inc., Gaithersburg, MD). Resulting cDNA was used with the ENZO BioArray High Yield RNA Transcript labeling kit (ENZO, Farmingdale, NY) to synthesize biotin-labeled cRNA; the latter was purified on an RNeasy spin column (QIAGEN) and chemically fragmented to a size range of 35 to 200 bp. cRNAs were concurrently hybridized to HG-U133A GeneChips (Affymetrix, Santa Clara, CA). Hybridizations were performed for 16 hours, followed by incubations with streptavidin-conjugated phycoerythrin, and polyclonal anti-streptavidin antibody coupled to phycoerythrin. GeneChips were scanned using an Agilent GeneArray laser scanner and images analyzed using Affymetrix MAS 5.0 software. Bacterial sequence-derived probes on the arrays served as external controls for hybridization, whereas the housekeeping genes β-actin and GAPDH served as endogenous controls and for monitoring the quality of the RNA targets. Unsupervised nearest-neighbor hierarchical clustering (Spotfire DecisionSite, Somerville, MA) identified a significant effect of culture condition/date of RNA collection on overall gene expression profiles. Therefore, to identify candidate KLF9 gene targets, the following was separately performed on the combinations of 4S/2AS and 9S/3AS. Intensity values of probe sets were imported into GeneSpring Gx 7.3 software for analysis. Values were processed using the Robust Multiarray Analysis algorithm for background adjustment, normalization and log2-transformation of perfect match values. Data were subjected to per-chip and per-gene normalization and analyzed for differences between cell lines (fold-change, S relative to AS, value of 1.3 or higher; P < 0.05, Student's t test). Transcripts that passed these filters for both the 4S/2AS and 9S/3AS cell line combinations comprised the final KLF9-regulated gene lists and were compared for gene overlaps. The final gene list (comprised only of overlapping transcripts) was annotated using NETAFFX [24] and NCBI Entrez [25]. The microarray data have been deposited in Gene Expression Omnibus [26] as series GSE11855.

Reproductive Biology and Endocrinology 2008, 6:66

cDNA microarray We used a custom-made utero-placental cDNA microarray that was developed in our laboratory as previously described [21,22]. In brief, a cDNA library was constructed from mRNA isolated from endometrial (caruncular and intercaruncular endometrium) and placental tissues (cotyledonary and intercotyledonary fetal membrane) of Japanese black cows. The PCR products of about 4,000 clones from the cDNA library were robotically spotted onto glass slides. The clones were simultaneously sequenced using the MegaBACE 1000 DNA Sequencing System (Amersham Pharmacia Biotech, Piscataway, NJ). The array contained 3,955 spots that were clustered into 1,738 unique genes on the basis of sequence analysis. An additional 35 genes that were not included in the cDNA library but had also been spotted onto the cDNA microarray were used for the analysis since these genes have been shown to be characteristically expressed during gestation in bovines and humans [4,11,19,23-27]. MMP related genes made up the bulk of this group. The cDNA microarray hybridization procedures were described in previous reports [21,22]. Briefly, two μg of poly (A)+ RNA were reverse transcribed in the presence of cyanine 3 (Cy3) or Cy5 fluorescence dye (Amersham Biosciences, Buckinghamshire, UK) using SuperScript II reverse transcriptase (Invitrogen) to make the hybridization probes. Identical samples were labeled separately with either Cy3- or Cy5-dye. Thus, two hybridization reactions could be carried out with the same sample. The arrays were sequentially washed with different concentrations of SSC solutions after 16 hr incubation at 65°C. The arrays were dried by centrifugation at 1,000 × g. Hybridization images were immediately scanned by a GenePix 4000B laser scanner (Axon Instrument, Union City, CA, USA) and analyzed with GenePix Pro 4.0 software. Data normalization was performed by previously described procedures [26,27]. The local background intensity of each array spot was smoothed by local weight regression (Lowess) and subtracted from the spot intensity data. The subtracted intensity data were subjected to non-parametric regression and local variance normalization since non-parametric regression can reduce intensity-dependent biases. The accuracy is improved over that of linear regression if the points in the scatter plot of Cy3 versus (vs.) Cy5 are not distributed around a straight line. Data for individual genes were obtained by averaging the intensity values of analogous spots on the microarray. Data were log2 transformed and used for cluster analysis. The Cluster 3.0 program was used for the hierarchical clustering. The hierarchically clustered data were visualized using the TreeView 0.99 program (M.B. Eisen's based clustering program [28],