This study describes a new chemometric tool for the analysis of chromatographic data: the Superposing Significant Interaction Rules (SSIR) is a variable selector coming from QSAR field that directly analyses the raw internal data coming from the chromatographic software. This allowed the identification of relevant volatile compounds in cork (treated and not treated samples in the industry) extracted by untargeted HS-SPME in a particular case for which traditional treatments (PCA, Discriminant Analysis) did not produced relevant results. The procedure has revealed the presence of compounds which are increased in the case of treated samples. The obtained classificatory model is robust, as it passed satisfactorily cross-validation tests (96% or more in performance for leave-one-out processes). This is the first time SSIR procedure is applied for the analysis of chromatographic information.
It is SSIR applicable to IR, NMR, spectorscopic signals, EEGG signals, etc.
thank you for your interest in our method. There is not a SSIR commercial program, but a collaboration (or cession of source code in Fortran) is feasible. Please, take into account that SSIR is a general procedure for variables selection or relationships discovering. It will be strongly depending on your field how the data is to be prepared and the code adapted. Note that in this article the application is for chemometrics. Other applications (with proper adaptations) are possible (e.g. in cancer classification, DNA analysis, QSAR, etc.)
Sincerely,
E.B.
