Matrix upload

Upload a matrix in a csv or RDS file

Maximun file size = 30Mb (~ 600 x 5000 matrix)

Click Demo
to load Corces-Buenrostro AML dataset

NMF params

Annotation upload

Upload a table with annotation data in a csv or RDS file

The first column should match the column names in the input matrix

Start NMF

Click 'Submit' button to start running NMF in the uploaded matrix and with selected parameters


ShinyButchR is an app to run Non-Negative Matrix Factorization (NMF) on a small input matrix using the Bratwurst package

You can explore the results in the NMF plots tab, or download the resulting H and W matrices in the save results tab.

ShinyButchR was developed by the Biomedical Genomics Group @ Health Data Science Unit at the BioQuant Center and Medical Faculty Heidelberg

Visit us here!

If you have any suggestion please create an issue in our GitHub repository hdsu-bioquant/ShinyButchR


Factorized uploaded matrix

NMF plots

Displays the optimal factorization rank metrics

Displays a heatmap of the H matrix for each factorization rank

Displays a riverplot of all factorization ranks

Save results

saves all the results as RDS objects

H Matrix Heatmap

H Matrix UMAP

Recovery Plots

Optimal factorization rank

Based on the results of the factorization quality metrics, the optimal number of signatures (k)
Minize the Frobenius error, the coefficient of variation and the mean Amari distance
Maximize the sum and mean silhouette width and the cophenic coefficient.

NMF riverplot

Riverplot or Sankey diagram showing the similarity between signatures across factorization ranks

Signature Specific Features Heatmap

Selected Signature Specific Features Heatmap

Download Signature Specific Features table

Select format

Save NMF object

Save NMF object to use later in R


This object is compatible with the latest version of ButchR

To install in R use:


Save H and W matrices

Save the H matrix or the W matrix in csv or RDS format