In this post, I document the complete process of installing and configuring the GEOS-Chem Classic model, including environmental setup, compilation, and runtime configuration. The log is based on my experience running GEOS-Chem on a Linux HPC system, and covers both the challenges I encountered and the solutions I implemented.
This post walks through a full FLEXPART v11 installation for ECMWF meteorological data with NetCDF and Fortran support, including resolving common compilation issues. All components were built from source on a Linux HPC system with user-level access.
With ultra-high resolution Orbitrap MS1 scan, I already obtain highly accurate chemical formulae. However, chemical information are still not enough.
First, each formula may correspond to multiple isomeric structures, and each isomer can carry unique information,such as its special emitters, formation pathways, or toxicity.
Even worse, some ions share nearly identical m/z values under the fitting threshold (20 ppm tolerance)but correspond to entirely different molecular formule.
Continuing from the previous post on molecular formula assignment, I encountered a common issue: overfitting during TOF peak assignment. In some cases, a narrow TOF m/z window was being assigned with too many candidate formulas, resulting in high uncertainty for each individual ion. In other words, we might obtain a wrong intensity for a correct formula, which affecting the dat reliability.
To address this, I implemented a refinement step after formula assignment:
Only the local maxima within a given ppm range are retained, along with known TOF apex peaks.
In this post, I document the full formula assignment pipeline using MFAssignR, tailored for Orbitrap-MS data post-peak list processing (There is no need to reinvent the wheel here).
The steps include quality control filtering, noise estimation via Kendrick Mass Defect (KMD) plots, isotopic filtering, recalibrant inspection, and final formula assignment.
In the previous post, I described the procedures for combining molecular information from both EESI-TOF and Orbitrap measurements. In our offline system, EESI-TOF can be regarded as semi-quantitative and more stable for long-term monitoring, making it ideal for tracking consistent ion trends. One key purpose of Orbitrap data is to guide the identification and fitting of TOF-detected ions, using its ultra-high resolution and accurate mass information.
However, in some cases, certain peaks are clearly detected by TOF but are either missing in Orbitrap or filtered out during the Orbitrap peak filtering and clustering process. These are often real chemical features, especially in complex mixtures, that may be weak or distorted in Orbitrap but still prominent in TOF.
To address this, I implemented a strategy to embed missing TOF peaks into the final clustered peak list, ensuring a more complete molecular representation, especially important in non-targeted or mixture-rich analyses.
After m/z calibration, the next step is to align ions detected across different retention times that likely originate from the same chemical species but appear at slightly shifted m/z values. To ensure robust clustering, it’s important to first remove rare or noisy ions that could interfere with pattern recognition.
After reading m/z data from Orbitrap raw files, it is essential to perform m/z calibration based on known reference masses of internal calibrants.
For long-term measurements, especially those spanning days or weeks, dynamic calibration is critical, since m/z drift can occur over time, and calibration parameters from Day 1 may not hold by Day 7. The function below allows automated chunk-wise recalibration across retention time.
The ultra-high resolution of Orbitrap mass spectrometry enables an unprecedented level of chemical detail. In my current research, we employ a direct-infusion Orbitrap approach (EESI inlet, Felipe et al, 2019) for non-targeted analysis. However, due to the lack of dedicated tools tailored for this application, I developed a new software package called OrbiTrack. This tool supports two core workflows:
(1) TOF peak guidance: Guiding ion fitting in EESI-TOF using high-resolution chemical information acquired at 120k resolution from Orbitrap.
(2) Directly utilization of Orbitrap MS1 data by integrating full time-series MS1 outputs, enabling online untargted molecular analysis without chromatographic sepeartion.
This blog post demonstrates how to plot FLEXPART model trajectories on a 3D terrain map, integrating topography and transport data for enhanced visualization. The workflow includes:
Update your browser to view this website correctly. Update my browser now