Collision cross section matching and annotation

collision-cross-section-matching-and-annotation

License: restricted — no clear open-source license detected for the underlying tool; verify licensing before commercial use or redistribution. <!-- asb-license-banner -->

Summary

Match experimental LC-IM-MS/MS 4D features (retention time, m/z, drift time, intensity) against a quantum chemistry calculation-assisted CCS prediction database to identify and annotate sterol isomers with structural detail (double-bond position, stereochemistry) and confidence scores.

When to use

When you have LC-IM-MS/MS data with measured collision cross section (CCS) values and m/z assignments, and you need to disambiguate sterol isomers (particularly N-Me derived unsaturated sterols) by matching against a curated database of predicted CCS values and MS/MS fragmentation patterns. Use this skill after 4D feature extraction but before final quantitative reporting.

When NOT to use

• Input data lacks drift time information or ion mobility separation (CCS matching requires IM-MS data, not LC-MS/MS alone)
• Sterol lipids have already been matched and annotated in a previous workflow step
• CCS prediction database is unavailable or not calibrated for the experimental instrument and sterol class

Inputs

• 4D feature table with retention time (RT), m/z, drift time (DT), and intensity
• Quantum chemistry calculation-assisted CCS prediction database (N-Me derived unsaturated sterols)
• Experimental MS/MS fragmentation spectra

Outputs

• Annotated sterol identifications with structure (double-bond position, stereochemistry)
• Matched feature records with CCS and spectral match quality scores
• Tissue-specific quantitative sterol data
• Confidence scores per identification

How to apply

Load the detected m/z and drift time values from the 4D feature extraction step. Look up each detected m/z in a quantum chemistry-assisted CCS prediction database of N-Me derived unsaturated sterol lipids to retrieve candidate structures and predicted CCS values. Filter candidate identifications by applying a CCS tolerance window (exact threshold from the paper's workflow); retain only matches within this window. Cross-validate the remaining candidates by comparing observed MS/MS fragmentation patterns against the database records using spectral similarity scoring. Assign confidence scores to each matched feature based on both CCS match quality and MS/MS spectral match quality. Annotate the final matches with sterol structure information (double-bond position and stereochemistry) and tissue-specific quantitative data.

Related tools

• Python (Primary programming language for implementing 4D feature matching, CCS tolerance filtering, and spectral similarity scoring logic)
• Jupyter Notebook (Interactive notebook environment for executing matching workflow steps, visualizing candidate matches, and generating annotated output tables) — github.com/Chen-micslab/QCCAssisted4DSterol
• scikit-learn (Provides SVR model and LASSO feature selection used in the underlying CCS prediction training; matching references predictions from this model)

Evaluation signals

• All matched features report both CCS match quality and MS/MS spectral match quality scores; no matches lack either component
• Matched sterol annotations include explicit double-bond position and stereochemistry; check that structure fields are not empty or generic
• CCS tolerance filtering is applied consistently: verify that all reported matches fall within the stated CCS tolerance window of the database
• Tissue-specific quantitative data is populated for each matched sterol; no quantitative records are NULL or missing for validated matches
• Confidence scores correlate with CCS and spectral match quality: matches with poor CCS alignment or low spectral similarity should receive lower scores

Limitations

• Workflow is optimized for N-Me derived unsaturated sterol lipids; applicability to other sterol modifications or lipid classes is untested
• Matching accuracy depends critically on the quality and completeness of the CCS prediction database; under-represented structural variants may not match or may be assigned to incorrect isomers
• CCS tolerance threshold must be tuned to the specific LC-IM-MS/MS instrument and calibration; the paper does not explicitly state the tolerance value used
• MS/MS fragmentation pattern matching requires clean, well-resolved spectra; noisy or contaminated spectra may fail to match or produce false identifications

Evidence

• [other] Match detected m/z values against a quantum chemistry calculation-assisted CCS prediction database of N-Me derived unsaturated sterol lipids: "Match detected m/z values against a quantum chemistry calculation-assisted CCS prediction database of N-Me derived unsaturated sterol lipids."
• [other] Filter candidate identifications by CCS tolerance and MS/MS fragmentation pattern matching: "Filter candidate identifications by CCS tolerance and MS/MS fragmentation pattern matching."
• [other] Annotate matched features with sterol structure (including double-bond position and stereochemistry) and assign confidence scores based on CCS and spectral match quality: "Annotate matched features with sterol structure (including double-bond position and stereochemistry) and assign confidence scores based on CCS and spectral match quality."
• [readme] LC-IM-MS/MS based 4D sterolomics data processing and identification: "LC-IM-MS/MS based 4D streolomics data processing and identification"
• [other] Load raw LC-IM-MS/MS data and apply 4D feature extraction (retention time, m/z, drift time, intensity) using Python: "Load raw LC-IM-MS/MS data and apply 4D feature extraction (retention time, m/z, drift time, intensity) using Python."