Mass spectrometry quality control: Instrument monitoring and pattern mining insights

Mass spectrometry quality control: Instrument monitoring and pattern mining insights Introduction Because of the inherent complexity of mass spectrometry, the results of an experiment can be subject to a large variability. In order to assess the quality of a mass spectrometry experiment, several different quality control metrics have been defined. These metrics can be computed from the experimental spectral-derived 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 parameters. This information provides complementary insights into the operational characteristics of a mass spectrometer compared to the aforementioned quality control metrics that originate from the mass spectral data. Methods The core of the instrument monitoring functionality consists of a central database that is optimized to store a vast amount of instrument parameters, called the iMonDB (Instrument MONitoring DataBase). Software tools exist to set up and populate the database in an automated fashion, and to visualize the instrument parameters stored in the iMonDB to perform a downstream analysis. Such structured recording of instrument parameters results in a large set of data detailing the instrument performance over time. Because of the high-dimensional nature of this data, advanced techniques are required to interpret it. For example, pattern mining algorithms that take into account the relationships between various parameters, such as subspace clustering, can be employed to analyze the high-dimensional data. Preliminary Data For each particular instrument the iMonDB contains the instrument parameters that were in effect during the execution of the experiments performed on the instrument. Possible parameters are for example the status of the ion source, the vacuum, or a turbo pump, depending on the type of instrument. These instrument parameters can be automatically extracted from experimental raw files and stored in the database to keep the iMonDB continuously up to date. Another type of information that is related to the operation of a mass spectrometer contained in the iMonDB are external events that may occur. For example, machine calibrations, periodic maintenance events, or even unexpected incidents that an operator likes to report. Unlike the instrument parameters, this type of information cannot be automatically retrieved and instead it has to be manually provided by the user. However, this information is vital when interpreting the evolution of the instrument parameters over time. It will be shown that the instrument parameters can be employed to detect/predict instrument failure. Monitoring of the instrument parameters showed that the FT turbo pump 4 of a Thermo Scientific Orbitrap Velos exhibited an increased power consumption. The reason for this was that while the turbo pump was deteriorating it tried to increase the power consumption to remain fully functional. However, this was insufficient to continuously achieve the required speed, after which the instrument finally broke down. By consistently monitoring the instrument parameters, the malfunctioning could have been predicted as the increase in power consumption serves as a proxy for turbo pump retrogression. Hence, by closely monitoring the instrument parameters and defining a normal range of operation, a suspected malfunctioning can be diagnosed early on and be reported. Based on this diagnosis, a timely intervention can prevent a highly undesirable loss of precious sample, analysis time, and effort. Novel Aspect Illustration how instrument monitoring can be used for quality control to detect instrument failure, introduction of instrument monitoring software