Table 1

Simulation studies of consensus module detection.

Noise level

Branch cut

Consensus module detection

Sensitivity

Specificity

Fidelity


1

0.965

1

1

0.989

1

0.975

1

1

0.988

1

0.985

1

1

0.985

1

0.995

1

1

0.965


2

0.965

0.966

1

0.964

2

0.975

0.984

1

0.958

2

0.985

0.998

1

0.949

2

0.995

1

1

0.935


3

0.965

0.717

1

0.871

3

0.975

0.823

1

0.838

3

0.985

0.929

1

0.824

3

0.995

0.997

0.999

0.822


4

0.965

0.457

1

0.823

4

0.975

0.589

1

0.744

4

0.985

0.739

0.997

0.713

4

0.995

0.928

0.995

0.675


5

0.965

0.0753

1

0.636

5

0.975

0.16

1

0.421

5

0.985

0.296

0.992

0.415

5

0.995

0.643

0.966

0.363


6

0.965

0.00345

1

0.667

6

0.975

0.0138

1

0.333

6

0.985

0.077

0.971

0.209

6

0.995

0.355

0.954

0.168


Using simulated data to assess the performance of the consensus module detection method. The column 'noise level' reflects the amount of noise added to the simulated data (details can be found in Additional File 7). The modules were defined as branches of an average linkage hierarchical cluster tree. The column 'branch cut' reports the heights used for cutting branches of the cluster tree. Sensitivity, specificity and fidelity are defined in the text.

Langfelder and Horvath BMC Systems Biology 2007 1:54   doi:10.1186/1752-0509-1-54

Open Data