« Previous - Version 15/18 (diff) - Next » - Current version
Ana Conesa, 05/26/2010 12:29 pm

On line example

Here you have a example of stress response in potato. The dataset contains 1000 genes, and 36 arrays. There are 3 stress conditions (heat, salt and cold) and one control. There are 3 time points and three biological replicates for experimental conditions.

Select your data

You can select here a file of the type Expression

Select variables

  • Continuous Variable Name. In this box you are prompted to select the continuous variable for the selection model, typically time or dosage. This variable has to be numerical.
  • Series Variable Name. Select here the variable that defines the different series in your dataset, typically treatment, strain, tissue, etc. This variable will be considered as categorical, even if it has a numerical nature.


  • Polynomial degree. The polynomial degree determines the complexity of the regression model. If degree is 1, data will be fit to a straight line. Degree = 2 allows for a curve; degree = 3 allows two inflexion points, and so on. The higher the value the more inflexion points in the expression profile one could consider. The maximum value is 1-number of time/dosage points. However high polynomial degrees are not recommended, specially if the replication is low, as data could be over-fitted. Values of 2 or 3 are good for more applications in short series (from 3 to 5 time/dosage points). If you have a long series, and want to find complicated patters, maSigPro could not be the best option and other methods such as ASCA-genes could be more appropriate.
  • Multiple testing adjustment. This is a significance adjustment when many genes are tested in the same analysis. The most common multiple testing procedure for microarray data is Benjamini and Hochber, 1995.
  • Significance Level for model variable. This refers the p.value imposed to variables in the regression fit to be retained in the model. This value influences the selection of genes differentially expressed between specific series comparisons. For example, for ControlVsTreatmentA comparison, or ControlvsTreatmentB comparisons, but does not affect the general assignment of genes as differentially expressed.
  • Cluster method. Once a list of differentially expressed genes is obtained for each series comparison, these genes are divided into groups of similar patterns for visualization purposes. Two cluster methods can be used to this end.


  • Job name. Give a short name to your analysis job
  • Job description. You can use this section to document further the characteristics of this analysis