Inferring Differentiation from Gene Expression

Ivan G. Costa, Stefan Roepcke, Christoph Hafemeister, Alexander Schliep

Software

A implementation of the Mixture of Trees and Mixture of Gausians are found here.

Code for Spectral clustering can be found at Spectral Lib. This code implements the method described in Ng. et al., On spectral clustering: Analysis and an algorithm. In NIPS, pages 849–856. MIT Press, 2001.

We used the SOM Toobox with SOM and k-means. NMF code can be obtained at NMF and SAMBA is implemented in the package Expander. In all these methods, the only parameter specified was the number of clusters: 5 for simulated data and 13 for the Lymphoid data. All other required parameters were set as default. As an exception, SAMBA does not require the number of clusters, as it finds it automatically (17 for the Lymphoid Data).

Data Sets

The pre-processed lymphoid data set is found in Lymphoid Development. The simulated data are found here SIM

Results

Web database, generated with our Mixture of Trees Report tool, with results of the analysis on the above data set is found in Lymphoid Development.