g

,

y

odel and a model of the mixture model algorithm, where x was a

the MF statistic for three replicates. All models were constructed

lusters only, where one cluster was for essential genes and the

s for non-essential genes.

eans(x,centers=2) # use K-means

eans(x,centers=2) # use Fuzzy c-means

lust(x,G=2) # use mixture model

umbers of predicted essential genes were 474, 474 and 435 using

eans model, the fuzzy C-means model and the model of the

model algorithm, respectively. Among them, 433 genes were

as essential genes by all three prediction models. The

nce rate was 90.97% (the highest) for 476 predicted essential

emonstrating the powerfulness of a multivariate model to

the uncertainty of gene essentiality analysis for this data set.

51 shows the Venn diagram of the consensus analysis of these

ate models.

The consensus analysis of the essential genes predicted by the K-means model,

-means model and the mixture model for three Francisella Tularensis replicates.

y

ve gene discovery is a subject for filtering out non-responsive

d discovering responsive genes for further investigation in an