General

We provide microarray-based gene expression profiling based on an "all Agilent" workflow. This service includes RNA labeling, hybridization, image acquisition, and data analysis. Prior to the initiation of any microarray experimentation (e.g. expression profiling), investigators must discuss experimental goals, design, expectations, RNA requirements, time of completion, and overall cost of the proposed experiment with microarray core personnel. Investigators will be provided with proven RNA extraction protocols and instructed to provide sufficient quantities of total RNA (currently 100 ng). Investigators outside of Marshall University are required to ship total RNA as a precipitate on dry ice by next day Fedex.

Experimental Design 

We recommend two types of basic experimental design: reference design and balanced block. In balanced block experiments, control and experimental RNAs are labeled with Cy3 and Cy5 dyes, pooled and hybridized to the same array. Balanced block design can be used only for simple class comparison type of analysis (e.g. finding genes that are upregulated or downregulated between two samples). In reference design, a commercial pool of RNAs is labeled with Cy3, combined with Cy5 labeled experimental RNA and hybridized to an array. In parallel, labeled universal RNA is combined with labeled control RNA and hybridized to a second array. This design is more desirable in cases where many comparisons will need to be made. In reference design experiments, the reference RNA should provide a signal for at least 80% of genes on the slide. We are currently using reference RNA from Agilent which is a collection of RNA pooled from several different cell lines or tissues. Human, mouse, and rat universal reference RNAs are available from several vendors. Reference design requires twice as many slides than balanced block design and is therefore more expensive.

Investigators may request that we perform standard and flip labeling for experiments (particularly for balanced block experiments) in order to eliminate dye bias associated with unequal incorporation of the cyanine dyes into cRNA.

Biological Replicates:

The required number of biological replicates (and hence the number of arrays) will vary with the type of experiment to be performed and the source of the RNA. For cell-line based experiments, investigators should plan on an absolute minimum of 6 biological replicates (8 preferred) for balanced-block experiments and an absolute minimum of 4 biological replicates (6 preferred) for universal reference designs. For inbred-animal model based experiments the number of replicates required may increase and for human population studies it can be much higher. In these cases microarray core facility staff can provide statistical analyses depending on the anticipated expression differences to be detected, desired false positive and negative rates, and data acquired in previous similar experiments in order to best advise the investigator. For most experiments, we recommend using one biological replicate per channel per array. Pooling may be necessary in cases where it is not possible to extract the required amount of RNA from a single sample.

RNA Quantity and Quality

RNA quality is the single most important factor in determining the outcome of your microarray analysis. The total RNA that you submit should be intact and free of protein and DNA contamination. Every RNA sample will be quality-tested by measuring the A260/A280 ratio and by nanochip analysis on the Agilent Bioanalyzer. Agilent has developed software that assigns a specific quality score (RNA Integrity Number or RIN) to the RNA sample based on its electrophoretic profile. The RIN ranges from 1 (totally degraded RNA) to 10 (completely intact RNA). Only high quality RNA, with RIN greater than 7 and A260/280 and A260/230 greater than 1.8, will be considered for microarray analysis. Column based RNA isolation methods are preferred since they provide RNA that is clear of organic contaminations. Presence of organic contaminations can compromise labeling efficiency.

Standard Labeling and Hybridization Protocol

For expression profiling, we employ the Agilent Version 4.0 protocol which relies on Agilent 60 mer microarrays in a 4 x 44,000 feature format, Low Input Linear Amplification for RNA labeling, improved hybridization and wash procedures, spike-in controls and Agilent Feature Extraction software v10.5.1.1 In the Linear Amplification RNA labeling chemistry [3], total RNA provided by investigators is initially converted into cDNA using an oligo (dT) primed reverse transcriptase reaction. This cDNA then serves as a template for cRNA synthesis by T7 RNA polymerase in the presence of cyanine 3- or cyanine 5-CTP. Linear amplification by T7 polymerase allows for a reduction in the amount of input total RNA from 20ug to as little as 100ng. This advance reduces the number of cultured cells needed for RNA extraction and the overall cost of the experiment. After column purification of labeled cRNA, equal amounts of cyanine 3- and cyanine 5- labeled cRNAs are combined and hybridized to the microarray at 65OC for 17 hours. Following hybridization, slides are washed and then scanned on an Agilent DNA Microarray Scanner. Preliminary fluorescence data will be extracted from each microarray and normalized using Feature Extraction software. Output from this stage of the analysis will include raw and normalized ratios presented in Excel spreadsheets and other representations.

Statistical Analysis of Microarray Data

Due to the large number of features on a microarray, sophisticated statistical analyses must be performed in order to be able to derive reliable conclusions from the data. In particular, great care must be taken to account for multiple-hypothesis testing in inferring statistically significant differences in expression between samples. Due to the interdependent nature of expression between genes, classical techniques for accounting for multiple-hypothesis testing tend to be too conservative and are not generally applicable to microarray data analysis. Several techniques have been specifically formulated to address these issues, and the core facility staff includes statisticians with expertise in analysis of microarray data. For class-comparison type experiments, we usually recommend using Significance Analysis of Microarrays (SAM, [1]), which is implemented by many pieces of specialized software including TIGR MeV. Other statistical analyses can be requested by the investigator, including pathway analysis, hierarchical clustering, k-means clustering, and Gene Set Enrichment Analysis [2]).

After completion of statistical analysis, investigators are invited to review their data set at a Core Lab Meeting (held weekly). This meeting gives statisticians the opportunity to explain the significance findings and criteria for significance (e.g. False Discovery Rates) and allow the investigator to ask questions.

Most journals now require that authors submitting manuscripts describing results of microarray experiments make the raw data and protocol descriptions available in MIAME-compliant format in public data repositories. For investigators wishing to submit such manuscripts, core facility staff will manage the submission of the data, provide links for potential reviewers, and manage public release of the data at the time of publication.

Microarray Slides and Cost of Expression Profiling

We currently use Agilent Microarray Slides for all gene expression profiling and keep human, mouse and rate whole genome arrays in stock. Other more specialized applications (e.g. microRNA expression profiling and ChIP-on-Chip) can be considered. The cost of a single expression profile (slide and reagents) is $400/array for Marshall University investigators. Other investigators should call for pricing.

Turn-around Time 

Given the time required for the completion of RNA analysis (one day), microarray labeling and hybridization (two days) and initial data analysis (2-3 days), the turn-around time will be at least 5 days from the time you submit your RNA samples, assuming that there is not a backlog of other requests.

Results of microarray experiment should be confirmed by another RNA quantification technique such as real-time PCR, Northern blot, or RNase protection assay. 

If you have any questions about the service or pricing, please contact:

Goran Boskovic, PhD
Microarray Core Facility Manager
E-mail: gboskovi@marshall.edu

US Mail Address:
Department of Biochemistry and Microbiology
Marshall University
Joan C. Edwards School of Medicine
1 John Marshall Drive
Huntington, WV 25704
Phone: (304) 696-7271
Fax:      (304) 696-7354