Themis: Batch Preprocessing for Ultrahigh-Resolution Mass Spectra of Complex Mixtures

Area of Science:

• Analytical Chemistry
• Computational Chemistry
• Petroleomics

Background:

• Fourier transform ion cyclotron resonance mass spectrometry (FT-ICR MS) provides high resolution for analyzing complex mixtures, such as petroleum.
• Current software struggles to efficiently process large datasets from multiple replicates, hindering data confidence and strategic decision-making.
• There is a need for improved data preprocessing methods to handle the increasing complexity and volume of FT-ICR MS data.

Purpose of the Study:

• To introduce Themis, a novel algorithm developed in R for joint preprocessing of replicate FT-ICR MS measurements.
• To enhance the consistency and reliability of data from complex mixtures by improving peak detection and reducing noise.
• To facilitate subsequent analysis by providing a robust list of reliably observed peaks across replicates.

Main Methods:

• Themis algorithm employs quality control criteria for detecting failed runs and ensuring comparable magnitudes across replicates.
• It includes peak alignment to standardize features across different measurements.
• An adaptive mixture model-based strategy is utilized to effectively distinguish true signals from noise.

Main Results:

• Themis preprocesses all replicate measurements in a single step, significantly improving data handling efficiency.
• It outputs a curated list of peaks consistently observed across replicates, enhancing confidence in the analytical results.
• The algorithm demonstrates improved consistency and reliability in peak identification for complex mixtures.

Conclusions:

• Themis provides a robust framework for preprocessing replicate FT-ICR MS data, particularly for complex samples like petroleum.
• The algorithm's ability to improve data consistency and peak reliability supports more confident downstream analysis.
• The flexible design of Themis makes it applicable to a wide range of complex sample analyses beyond petroleomics.