Analysis of mass spectrometry data of cerebral stroke samples: an evolutionary computation approach to resolve and quantify peptide peaks

A preliminary investigation of cerebral stroke samples injected into a mass spectrometer is performed from an evolutionary computation perspective. The detection and resolution of peptide peaks is pursued for the purpose of automatically and accurately determining unlabeled peptide quantities. A theoretical peptide peak model is proposed and a series of experiments are then pursued (most within a distributed computing environment) along with a data preprocessing strategy that includes (i) a deisotoping step followed by (ii) a peak picking procedure, followed by (iii) a series of evolutionary computation experiments oriented towards the investigation of their capability for achieving the aforementioned goal. Results from four different genetic algorithms (GA) and one differential evolution (DE) algorithm are reported with respect to their ability to find solutions that fit within the framework of the presented theoretical peptide peak model. Both unconstrained and constrained (as determined by a course grained preprocessing stage) solution space experiments are performed for both types of evolutionary algorithms. Good preliminary results are obtained.