The corresponding programming codes for computing a function concept consistency score are available upon a written request to the authors.
Ideally, we should consider the hierarchical structure of GO when we computed concept consistency between two nodes.
In order to identify tumour subtypes both biologically meaningful and clinically relevant, we evaluated a functional concept consistency score for a gene subset to define its biological meanings.
In order to establish the functional validation for the identified gene subsets, we defined a GeneOntology (GO -based [ 19- 21] conceptual functional similarity measure, called concept consistency score (see Methods for detail).
Therefore, in this study, we designed a functional consensus score for evaluating a candidate gene subset in terms of functional concept consistency, which is similar to the biological homogeneity index (BHI) proposed recently [ 43].
The steps for computing a concept consistency score [ 49] for gene subsets G i (i = 1, 2,..., N) were described as follows: For g j ∈ G i, we mapped gene g j to a node(s) of GO via the links between three databases: GeneBank, Unigene, and LocusLink.
This concept combines the concept of consistency with the concepts of attributes and closeness.
New concepts of consistency, generalized monotonicity, and stability are introduced; among them, the generalized monotonicity and consistency, which are easier to verify in practice, are natural extensions of the corresponding notions of finite difference methods for first order fully nonlinear Hamilton Jacobi equations.
The validity of the analysis must be judged with criteria appropriate for the method, namely proper use of concepts, logical consistency, internal coherence and plausibility.
