The RRHO algorithm allows for the comparison of two gene expression signatures. Each signature is processed as a ranked list based on expression differences between two classes of samples. The signatures can be input either as raw expression data and sample and class labels, or as a preranked gene list.Getting started
This is optional if both data sets are already using common ID's. The annotation file is tab-delimited text file with each line having the columns listed below. A header row is not needed.
When the annotations are used, the identifiers in the GCT file will be converted to a UniGene ID. Multiple genes having the same UniGene ID will be collapsed according to the highest t-test p-value between the two classes.
If the two datasets are from different species, UniGene annotations are required so that homologs between genes can be found.
Annotations for several commonly used Affymetrix GeneChips are provided.
This version of Rank Rank Hypergeometric Overlap (RRHO) currently recognizes only Mus musculus and Homo sapiens.
Rank rank options
The step size is used to bin the ranked items to improve the run time of calculating the hypergeometric distribution. An optimal step size is small enough to match the image resolution to the overlap patterns in the data, but large enough to reduce computational time. For gene expression data on the scale of 10,000-50,000 probes, we recommend a step size of 100-500.
Reverse rank lists will reverse the ranks of the items in one or both of the datasets.
In addition to the hypergeometric heat map, you may select two additional graphs:
Benjamini-Yekutieli multiple hypothesis correction can be applied.
The output can include a list(s) of the overlapping genes within the region(s) of maximal statistical overlap.
The following parameters are used as an example analysis of Rank Rank Hypergeometric Overlap (RRHO). Feel free to download the sample files and try it yourself.Show sample files and parameters [+]
A simple version of Rank Rank Hypergeometric Overlap (RRHO) is available if you already have a list of ranked items that you want analyzed.