Markov-chain-based heteroscedastic regression model for the analysis of high-resolution enzymatically 18O-labeled mass spectra.

The enzymatic (18)O-labeling is a useful technique for reducing the influence of the between-spectrum variability on the results of mass-spectrometry experiments. A limitation of the technique is the possibility of an incomplete labeling, which may result in biased estimates of the relative peptide abundance. We propose a Markov-chain-based regression model with heterogeneous residual variance, which corrects for the possible bias. Our method does not require extra experimental steps, as proposed in the approaches proposed previously in the literature. On the other hand, it includes some of the alternative approaches as a special case. Moreover, our modeling approach offers additional advantages over the previously developed methods, including the possibility of the analysis of multiple technical replicates for samples from different biological conditions, with an assessment of the between-spectra and biological variability.