View the clustered data
View figures from the paper
Additional information
Get help interpreting
the data  
Download Data
Download Software

Help using FuzzyK

The input of the FuzzyK program is a text tab-delimited file in .cdt format.

To run FuzzyK, call the program, load the input .cdt file, and run FuzzyK with the appropriate parameters. 1. Call the Aerie program
   > ./aerie
2. Load the input .cdt file
   Aerie> load filname.cdt
   Loaded 100 rows
   Loaded 200 rows 
   Loaded 300 rows
   Loaded 400 rows
   Loaded 500 rows
   Loaded 600 rows
   Loaded 700 rows
   Loaded 800 rows
   Done Loading filename.cdt
3. To run FuzzyK with extra features, enter the paramters in the correct order (with either the correct variable or a "0" for null)
 Aerie> fuzzy 50 g e s a
  • "fuzzy" calls FuzzyK
  • "50" is the total number of clusters (k) requested
  • "g" = calculate gene weights to underweight genes in the first clustering round
  • "e" = calculate experiment weights to underweight experiments in the first clustering round
    (see Cluster Manual for details on the gene and experiment weighting functions)
  • "s" to seed prototype centroids with user-defined gene expression patterns
  • "a" to add user-defined gene expression patterns (ie. user-defined centroids) to the final list of unique centroids

    You can use subsets of these paramters by entering "0" for features that are not requested
       Aerie> fuzzy 60 g e 0 0 
    You can also run the basic FuzzyK by simply entering the first two parameters
       Aerie> fuzzy 120 
    4. The program will ask you to enter a file name to save the data. Depending on the parameters entered, the program will also request other details from the user.

    5. The program will generate two files, both in .cdt format.
  • The first file contains the unique centroids identified by FuzzyK and has the suffix ".fct"
  • The second file contains the gene-centroid memberships and has the suffix ".mb"

    These files, in addition to the dataset on which the program was run, are required to view the data in FuzzyExplorer.

    6. Note: sometimes the gsl SVD function used to compute the Eigen vector prototype centroids cannot compute the principal components - in this case, a gsl error will appear on the screen. A simple solution is to rerun FuzzyK with a slightly different k. This problem appears to have been largely corrected in the new SVD function in the latest GNU scientific library, with which the available FuzzyK software was compiled.