Computational mass spectrometry of linear binary synthetic copolymers

Abstract

The accurate characterization of synthetic polymer sequences represents a major challenge in polymer science. We present a computational approach to quantify the abundances of all sequences in a measured copolymer sample. The first step in our workflow is transforming mass spectra into copolymer fingerprints. Our method is based on linear programming and is capable of automatically resolving overlapping isotopes and isobaric ions. Peak intensities in matrix-assisted laser desorption/ionization spectra are influenced by mass and composition-dependent ionization. We demonstrate a method to correct the abundance bias. The second step in our workflow is interpreting the computed copolymer fingerprints using new Markov chain models for copolymerization kinetics: The Bernoulli and Geometric models. In contrast to previous Markov chain approaches to copolymerization, both models take variable chain lengths and time-dependent monomer probabilities into account and allow computing sequence likelihoods and copolymer fingerprints. We find that computing the models is fast and memory efficient. Then, we focus on the Geometric copolymerization model with reactivity parameters. First, several approaches to identify the optimal model parameters from observed fingerprints are evaluated using Monte-Carlo simulated data. A compromise between robustness and running time is found by exploiting the relationship between ordinary differential equations and the Geometric model. Second, we show that the model is also useful for copolymerizations involving termination and depropagation reactions. We then compute several copolymer statistics and compared them to the statistics obtained from Monte-Carlo simulations. Last but not least, we present our software framework COCONUT, which implements all algorithms presented in this thesis. Our software is freely available and provides a graphical user interface. COCONUT represents a step towards comprehensive computational support in polymer science.