Are there alternatives to MCR-ALS for determining pure response profiles of a mixture? MCR-ALS says it is a tool for "the recovery of the pure response profiles of the chemical constituents or species of an unresolved mixture when no prior information is available about the nature and composition of these mixtures."  Glotaran/TIMP also offer singular value decomposition.  Are there other tools I could consider using for similar purposes?
I am particularly interested in comparing the tolerance of tools to missing data and the quality of the documentation.
 A: You could consider the Matlab N-way toolbox by Rasmus Bro et al or the commercial version, the PLS Toolbox by Eigenvector Research, NMFLab or NTFLab for Matlab, MVC2 or MVC3, or the osd,  ALS, NMF, multiway or metaMS packages in R. I've tried most of these for separating coeluting compounds in GC/MS chromatograms but wasn't all that impressed by any of them unfortunately - most need quite a lot of tinkering to make them work, and imposing a sufficient number of constraints is really critical to get good results (e.g. nonnegativity, unimodal peak shapes, good initial estimates, correct estimation of chemical rank/correct nr of peaks etc) - and most are not robust enough yet for routine research. But it can be done! I was just working on a modified MCR-ALS algorithm that uses elastic net regularization to enforce sparseness plus unimodality and nonnegativity constraints that seems to perform reasonably well... 
Good reviews are Cichocki et al.'s book "Nonnegative matrix and tensor factorizations: applications to explanatory multi-way data analysis and blind source separation" and the "Handbook of Blind Source separation: Independent Component Analysis and Applications". There is also lots of software for specific applications, e.g. for GC/MS data the free program AMDIS, or the commercial ChromaTOF or Scanalytics Analyzer Pro.
