Journal watch: AI-empowered data-independent acquisition mass spectrometry-based quality control

Open Access:
Nature Communications

SYNOPSIS

The study published in Nature Communications presents a comprehensive investigation into quality control (QC) practices for mass spectrometry (MS)-based proteomics, with a particular focus on data-independent acquisition (DIA) workflows. As quantitative proteomics scales up, ensuring the reliability of liquid chromatography tandem mass spectrometry (LC-MS/MS) workflows is crucial to achieving precise protein identification and quantification.

Traditionally, QC in proteomics relies on periodic analysis of standardised samples in data-dependent acquisition (DDA) mode, assessing performance through metrics such as the number of peptide and protein identifications. However, this approach lacks the sensitivity to detect subtle changes in LC-MS performance, especially as data acquisition techniques evolve. This paper addresses these limitations by systematically comparing DIA and DDA QC metrics across diverse experimental conditions, ultimately establishing a robust framework for DIA-based QC.

The researchers gathered an extensive dataset comprising 2,754 DIA files and 2,638 paired DDA files from mouse liver digests, analysed using 21 mass spectrometers of eight different models across nine laboratories over a period of 31 months. This large-scale, multivendor dataset provided a unique opportunity to rigorously assess the performance of DIA-based QC metrics in detecting changes in LC-MS status.

Central to the study was the role of liquid chromatography, which underpins the separation of complex peptide mixtures prior to mass spectrometric analysis. Variations in chromatographic performance can profoundly impact the accuracy and reproducibility of protein quantification. The study highlights that DIA-based QC exhibited greater sensitivity in capturing these variations compared to DDA-based approaches, providing a more nuanced evaluation of LC-MS performance.

A key contribution of the work lies in the prioritisation of 15 consensus QC metrics tailored for DIA workflows. These metrics encompass aspects related to chromatography, ion source stability, MS1 and MS2 performance, dynamic sampling, and peptide identification. To ensure the robustness of these metrics, the authors recruited 21 experts from 10 laboratories to manually inspect the 2,754 DIA QC files.

The expert evaluations served as the foundation for training an artificial intelligence (AI) model aimed at automating DIA-based QC assessments. The model achieved impressive area under the curve (AUC) values of 0.91 for LC performance and 0.97 for MS performance on the first validation dataset, with corresponding values of 0.78 and 0.94 on an independent validation set, demonstrating its capacity to reliably detect changes in LC and MS status.

One of the most significant aspects of the study is its focus on the interplay between liquid chromatography and mass spectrometry. In high-throughput proteomics, chromatographic stability directly impacts the reproducibility of peptide ionisation, retention times and overall MS signal quality. By enhancing QC metrics to better monitor LC performance, the study addresses a longstanding gap in proteomics workflows, where deviations in chromatographic conditions often go unnoticed until they manifest as broader declines in protein identifications. The ability of DIA-based metrics to more sensitively capture these deviations underscores the importance of robust chromatography in large-scale proteomics experiments.

In summary, this work advances the field of proteomics QC by shifting focus towards DIA workflows and emphasising the critical role of liquid chromatography in ensuring experimental reproducibility. The integration of expert-driven consensus metrics with AI-powered assessment offers a robust and scalable solution to the growing demands of high-throughput proteomics. As proteomics continues to evolve, the methodologies presented here promise to enhance data quality and provide deeper insights into the intricate relationship between chromatography and mass spectrometry performance.

https://doi.org/10.1038/s41467-024-54871-1

http://creativecommons.org/licenses/by-nc-nd/4.0/