Constrained Mixture Estimation for Analysis and Robust Classification of Clinical Time Series

Ivan G. Costa, Alexander Schoenhut, Christoph Hafemeister,Alexander Schliep


The method description can be find in the publication

Costa, I. G., Schonhuth, A., Hafemeister, C., Schliep, A. Constrained mixture estimation for analysis and robust classification of clinical time series. Bioinformatics (Oxford). , v.25, p.i6 - i14, 2009. Paper.


Scripts for running the Constraint based Mixture Estimation are found here. It requires the installation of the GQL 2.0 (unreleased version) GQL software and of Pymix.

Data Sets

The pre-processed log MS data set is found in MS data. First column indicate patients, second the class (1 for good responders 0 for bad responders), subsequent columns are the measurements of the 70 genes for 7 time points. We also provide the data with the labels and constraints used in the cross-validation Constraints and Labels.

The simulated data are here for original Lin et al. data; here for data with added noise and here for data with mislabeled patients. It follows a similar format as before, now with 100 genes at 8 time points.