Skip to main content
. 2014 Mar 20;9(3):e90801. doi: 10.1371/journal.pone.0090801

Table 3. Various specifications considered for the biclustering methods.

Specifications Description
GOVL The obtained biclusters are allowed to have overlaps over only the gene-sets.
SOVL The obtained biclusters are allowed to have overlaps over only the sample-sets.
GSOVL The obtained biclusters are allowed to have overlaps over both gene and sample-sets.
NOVL No overlaps at all are allowed for the obtained biclusters.
DISC Discretization is mandatory for running the algorithm
TMV The method is tolerant to missing values.
SIMA Simulated annealing is applied to avoid convergence to local optima.
VECOS Vector Cosine Scores is applied to measure the similarities of the samples (or genes).
SVD The method applies a form of Singular Value Decomposition.
MCMC The method employs a Markovian Chain Monte Carlo approach.
BAYES The method employs a fully Bayesian approach.
EM The method uses the Expectation-Maximization method.
MOTIF The MOTIF sequence co-occurrence is considered in the biclustering approach.
TREE The method applies a tree structure for discovering suitable sets of genes and samples.