Visunet: Graph visualization for classification rules

VisuNet: Accepted Rules File Format

There are two formats for rules used by VisuNet as of now: Rosetta and Line by Line.

Rosetta File Format

This format is produced by the Rosetta program. For each rule, there are six lines. The first line is the rule itself in the form:
        IF <feature1(value1)> [AND <feature2(value2)>...] => <decision1(value1)> [ OR <decision2(value2)>...]

        e.g. IF sunny(yes) AND windy(yes) => play(no)
The next lines are rule statistics in the form of:
      Supp. (LHS) = [1 object(s)]
      Supp. (RHS) = [1 object(s)]
      Acc.  (RHS) = [1]
      Cov.  (LHS) = [0.0042735]
      Cov.  (RHS) = [0.0357143]
      Stab. (LHS) = [1]
      Stab. (RHS) = [1]
      Supp - Support
      LHS/RHS - Left Hand Side/Right Hand Side of rule
      Acc - Accuracy
      Cov - Coverage
      Stab - Stability

Line by Line Format

Each rule is represented in a single tab-separated line as:
      <feature1(value1)>,[<feature2(value2)>...]  <decision1(value1)>,[<decision2(value2)>...]  <accuracy>  <support>
        sunny(yes),windy(yes)  play(no) 3 0.75
The condition list (feature=value) and the decision list (decision=value) columns cannot have commas in the text otherwise the output may be gobbled

Color of Nodes Format

Accuracy Value Format

Use Rosetta or Line by Line file formats.

Level of the Gene Expression Format

Use the Line by Line file format. Gene expression levels should be represented as integers:
        p53(1),hsf1(2)  cancer(0) 3 0.75
        p53=1,hsf1=2  cancer=0 3 0.75
or string value:
        p53(low),hsf1(high)  cancer(0) 3 0.75
        p53=low,hsf1=high  cancer=0 3 0.75

Mapping File Format

The mapping file is a two-column tab-separated file containing, on each line, a condition (feature=value) and a gene short name (e.g. BRCA1) respectively. This file is used to populate the network nodes with KEGG and GO data by matching a feature to a gene it represents and is therefore only useful if you have genomic data. This includes microarray data, CpG sites, Histone Methylations that can be mapped to specific genes.