|
Expression Analysis provides statistical and bioinformatic data analysis services that help explain the large amounts of microarray data commonly generated by gene expression and genotyping experiments. The analysis often begins with an experimental design consultation to discuss important aspects of population variation, genomic coverage, and expected results. Expression Analysis has developed useful standard methods to facilitate projects. When necessary, Expression Analysis can develop and deliver custom solutions.
Bioinformatics Services for Expression Assays
Bioinformatics Services for Genotyping Assays
Other general bioinformatics topics are discussed below.
Experimental Design and Consultations
Replication
Novel Analysis Methods
Experimental Design and Consultations
A successful microarray research project starts with a well-designed experiment. Expression Analysis frequently provides consultations with our Ph.D. statisticians and/or scientists to discuss experimental design. After data transfer, each investigator can request a completion call from one of our Ph.D. scientists to review interesting results or ask about the potential for follow-on studies.
When we consider your experiment, important issues include:
Specification of goals: Is the primary interest in gene discovery, class discovery or class prediction?
Breadth of objectives: What situations or factors should the experiment include to be representative?
Sources of variability: What are the possible causes and how do we control unwanted variation?
Sources of bias:
What are the possible causes and how do we control unwanted bias?
We commonly advise clients regarding many types of studies, including:
Two-group comparisons: Baseline vs. Experimental groups
Multi-group comparisons: Baseline vs. Treatment Group A vs. Treatment Group B
Time-course experiments with repeated measures of the same subject
Complex multi-way experiments involving two or more factors that have individual or synergistic effects
Predictive marker experiments that associate expression values to complex of continuous phenotypes
Back to top
Replication within Experiments
Many clients request advice on the number of samples to be included in microarray studies. Expression Analysis recommends that for statistical inferences, at least three biological replicates are needed for any study, but five or more replicates are preferred. As population variances increase, so do the number of replicates required to adequately query the population. For example, human studies usually require at least twice as many samples as animal studies. Blood samples typically require more replicates than other tissue types. In contrast, studies with cell lines require only a few replicates to ensure assay reproducibility.
Expression Analysis recommends the inclusion of biological replicates, rather than technical replicates. We find that the high quality of GeneChip manufacturing and sample processing obviates the need, in general, for purely technical replicates. Biological variability is typically much greater than technical variability. Precious resources are much better leveraged by including more biological replicates in the study, if available.
Back to top
Novel Analysis Methods
Expression Analysis has developed novel analysis methods for identifying more differentially expressed genes and for predicting “No Call” genotypes in population studies. These methods are a result of our extensive expertise with the Affymetrix platform and reflect our commitment to provide a higher level of service to our clients.
Back to top
|
