bioNMF Logo

bioNMF Results


bioNMF 2.0 - Main Menu

Job's parameters:
Job's ID: Biclustering_example
Analysis method: Biclustering analysis
Input file format: ASCII-text
Data matrix has numeric column-headers: No
Data matrix has numeric row-labels: No
Transpose data matrix: No
Normalization method: Do not normalize
Method to make data positive: Exponential scaling
Initial values of output matrices W and H: Selected randomly
NMF algorithm: Non-smooth NMF
Range of factorization ranks: [2...5]
Number of runs per rank: 40
Number of iterations per run: 2000
Stopping threshold: 40 iteration(s)
Save option: Combine matrices in a single file

NMF results:
Best factorization rank:4
Best run (0-based indexing):4
Distance between input matrix and W*H: 1.383587e-05
Cophenetic Correlation Coefficient(s)
Heatmap of input matrix Profile plot of input matrix
   
Matrix W Matrix H
Matrix W (numeric data) Matrix H (numeric data)
Heatmap of matrix W Heatmap of matrix H
Profile plot of matrix W Profile plot of matrix H

Biclustering Analysis results:
Number of biclusters found:   4
Results shown for each bicluster found:
  • A heatmap (bottom) with the subset of genes and all samples sorted by its association to the local pattern.
  • A plot (top) representing the coefficients of all samples in the corresponding row of matrix H. Marked in blue are samples that show the largest coefficient for that factor, while in green are marked samples that show largest coefficients in other rows in matrix H.
  • The consistency value represents the percentage of times that this bicluster has been found in 40 run(s) of the algorithm (see number of runs per rank parameter). We consider that two biclusters are the same if they share 80% of the genes.
  • Bicluster submatrix
  • Row indexes of the extracted bicluster.
  • Column indexes of the extracted bicluster.
More details can be found in Carmona-Saez et al., BMC Bioinformatics, 2006.
Bicluster #1:

Bicluster data
Row indexes (1-based indexing)
Column indexes (1-based indexing)

Bicluster #2:

Bicluster data
Row indexes (1-based indexing)
Column indexes (1-based indexing)

Bicluster #3:

Bicluster data
Row indexes (1-based indexing)
Column indexes (1-based indexing)

Bicluster #4:

Bicluster data
Row indexes (1-based indexing)
Column indexes (1-based indexing)

Supplemental matrices:
Combined matrices W Combined matrices H
Combined matrices W (numeric data) Combined matrices H (numeric data)

Output log (detailed information and timings).

bioNMF 2.0 - Main Menu