A comprehensive approach to mass spectrometry quality control

Introduction

Because of the inherent complexity of mass spectrometry, the results of an experiment can be subject to a large variability. As a means of quality control, several qualitative metrics have been defined. These quality control metrics can be derived from the experimental data, and they capture the important operational characteristics of a mass spectrometer.

Additionally, besides these metrics based on the experimental data, there is a separate part of qualitative information available in the form of the mass spectrometer instrument settings. This information can provide complementary insights into the operational characteristics of a mass spectrometer compared to the aforementioned quality control metrics.

Instrument settings

Several tools exist to compute quality control metrics based on the experimental data, however, unfortunately, until now the instrument settings have been largely ignored. In fact, this potentially important source of quality control information is still largely untapped.

Therefore, we have developed tools to extract instrument settings from experimental raw files, and store this qualitative information in an optimized relational database. This allows for monitoring historical instrument settings for a large range of experiments over the course of a considerable time period in order to detect and confirm instrument failure, fostering additional methods of quality control.

Pattern mining

Both the experimental-based quality control metrics and the instrument settings can be used to monitor instrument behavior, in order to detect emerging problems in a timely fashion. Previous approaches mostly evaluated each metric individually; however, looking at specific metrics individually is insufficient because the different stages of a mass spectrometry experiment do not function in isolation, instead they can have an effect an each other. Therefore, specific pattern mining algorithms, which aim to take into account the relationship between various metrics, can be employed to provide additional insights when analyzing quality control data. In particular, subspace clustering can be used to find subspaces of related metrics, based on the experimental data as well as the instrument settings, and provide a comprehensive analysis of various different sources of qualitative data.

Conclusion

The awareness has risen that suitable quality control information is mandatory to assess the validity of a mass spectrometry experiment. Over the past few years, effort has been expended to define a suitable set of qualitative metrics based on the experimental results. We have complemented this source of information with the mass spectrometer instrument settings, which provide complementary insights into the operational characteristics of a mass spectrometer. Furthermore, we have developed specialized pattern mining algorithms to interpret this high-dimensional data. The derived patterns could subsequently be used to optimize experimental design and mass spectrometry instrument settings.