Development of smart acquisition methods using real-time control of instruments and machine learning
Supervision
Odile Schiltz
IPBS, 1st Supervisor
Mathias Wilhelm
TUM, 2nd Supervisor
Objectives
Develop intelligent data acquisition workflows to enhance proteomic analytical depth. While global MS acquisition methods based on DDA enable the direct, large-scale and unbiased analysis of complex samples, low-abundant proteins remain undetected in such single-run measurements. Approaches such as PRM or DIA take advantage of previously acquired information (e.g. validated peptide sequence matches, libraries…) to improve the detection of low-abundant molecules during analysis of the samples. The problem of retention time (RT) variability complicates the efficient use of previously collected data when programming scheduling methods to specifically detect at high sensitivity many low abundant proteins (high-throughput PRM). We plan to develop new open-source software solutions for real-time control of MS acquisition in order to improve the implementation and multiplexing of high sensitivity PRM measurements, based on optimized RT prediction.
Methodology
The ESR will develop new open-source software solutions for real-time control of instruments, integrated with the MS-Angel tool for the automation of data processing tasks and subsequent triggering of customized acquisition methods. He will also evaluate machine learning algorithms and existing solutions (e.g. Prosit) to improve the parameters of peptides fragmentation.
Required Skills
Knowledge of data acquisition by mass spectrometry
C# programming language (experimented)
Expected Results
An open-source framework facilitating the development of new real-time and smart acquisition methods, improving the throughput of PRM assays.
Planned Secondments
Host: TUM (M. Wilhelm), Duration: 3 Months; When: Year 2; Goal: Methods optimization with Prosit.
Host: EKUT (O. Kohlbacher), Duration: 1 Month; When: Year 3; Goal: Integration with OpenMS.
Enrolment in doctoral programs
Ph.D. in Bioinformatics from University of Toulouse.
References
1. Bouyssié, D., Hesse, A. M., Mouton-Barbosa, E., Rompais, M., Macron, C., Carapito, C., Gonzalez de Peredo, A., Couté, Y., Dupierris, V., Burel, A., Menetrey, J. P., Kalaitzakis, A., Poisat, J., Romdhani, A., Burlet-Schiltz, O., Cianférani, S., Garin, J., & Bruley, C. (2020). Proline: an efficient and user-friendly software suite for large-scale proteomics. Bioinformatics (Oxford, England), 36(10), 3148–3155. https://doi.org/10.1093/bioinformatics/btaa118
2. Gessulat, S., Schmidt, T., Zolg, D. P., Samaras, P., Schnatbaum, K., Zerweck, J., Knaute, T., Rechenberger, J., Delanghe, B., Huhmer, A., Reimer, U., Ehrlich, H. C., Aiche, S., Kuster, B., & Wilhelm, M. (2019). Prosit: proteome-wide prediction of peptide tandem mass spectra by deep learning. Nature methods, 16(6), 509–518. https://doi.org/10.1038/s41592-019-0426-7