Chemical Space Coverage in LC-HRMS: Selectivity and Bias

Investigating the Chemical Space Coverage of Liquid Chromatographic Techniques

MSc Chemistry Thesis - Analytical Sciences | Dec 2025

---

Photo by Ryan Zazueta on Unsplash

Thesis Presentation: Chemical Space Coverage in LC-HRMS

Investigating the chemical space coverage of liquid chromatographic techniques: how selectivity drives measurability

---

Photo by Nastuh Abootalebi on Unsplash

Literature Thesis Presentation

Investigating the chemical space coverage of liquid chromatographic techniques: how selectivity drives measurability

---

Photo by Nastuh Abootalebi on Unsplash

• Introduction to Chemical Space and NTA
• Methodology: RepoRT Repository
• Analysis of Chemical Coverage
• Trends in Reported Compounds
• Conclusions & Future Perspectives

---

Photo by Nastuh Abootalebi on Unsplash

• Introduction and Research Goal
• Methodology and Data Acquisition
• Analysis of Chemical Space Coverage
• Key Trends and Findings
• Conclusion and Future Perspectives

---

Photo by Bench Accounting on Unsplash

• Introduction: Chemical Space and NTA
• Methodology: RepoRT Analysis
• Analysis of Chemical Coverage
• Key Findings and Trends
• Conclusions and Future Perspectives

---

Photo by Tom Parkes on Unsplash

1

Introduction

Understanding the chemical space of the exposome and the role of LC-HRMS in non-targeted analysis (NTA).

---

Photo by Tom Parkes on Unsplash

1

Introduction & Research Goal

Understanding the Exposome and Selectivity-Measurability Bias

---

Photo by Vedrana Filipović on Unsplash

• Exposome encompasses totality of chemical exposures over a lifetime.
• Non-targeted analysis (NTA) with LC-HRMS is essential for identifying thousands of unknown compounds.
• Selectivity-measurability bias: analytical methods only detect compounds interacting with the system.
• Objective: Investigate how different LC conditions capture various regions of chemical space using the RepoRT repository.

• The exposome encompasses the totality of lifetime exposures.
• Characterization requires analytical methods detecting thousands of diverse chemical entities.
• Non-targeted analysis (NTA) with LC-HRMS is essential but constrained by methodological selectivity.
• Selectivity-measurability bias: Only chemicals interacting with the system are measured, potentially missing classes of compounds.
• Research Goal: Evaluate the chemical space coverage of 236 curated LC methods using the RepoRT repository.

• The exposome consists of all chemical/non-chemical exposures over a lifetime.
• Chemical space refers to all plausible organic structures relevant to human/environmental health.
• Non-Targeted Analysis (NTA) + LC-HRMS is essential for identifying unknown exposures.
• Selectivity-measurability bias: Only chemicals interacting with the analytical system are measured.
• Small parameter changes in LC workflows can lead to significant differences in detected compounds.

2

Methodology

Data acquisition and RepoRT database setup

---

Photo by Deng Xiang on Unsplash

2

Methodology

Data acquisition from the RepoRT repository and experimental setup.

---

Photo by Tom Parkes on Unsplash

• 236: LC Methods
• 17,083: Unique Compounds
• 89%: RPLC Dominance
• 11%: HILIC Representation

• RepoRT repository: 236 curated LC setups processed using Julia.
• Metadata: Filtered for specific column parameters (name, USP code, length, etc.).
• Compound data: 75,797 compounds retrieved with chemical descriptors (TPSA, Log-P, Mass).
• USP classification: 89% RPLC (primarily L1), 11% HILIC (primarily L122).
• Normalization: Retention times normalized on a 0-1 scale to allow comparison across different methods.

• Utilized 236 curated LC methods from the RepoRT repository.
• Instrumental setup metadata filtered for consistency (column name, length, temperature, flowrate, etc.).
• Data processed with Julia 1.11.4; descriptors retrieved via PubChemCrawler.jl.
• Analytes characterized by exact mass, predicted log-P, H-bond properties, and TPSA.

• Chemical space coverage is heavily biased towards RPLC conditions.
• Measured compounds range widely in theory (TPSA 0-852, log-P -11 to 26), but are constrained in practice.
• Most frequently measured region: TPSA 0-200, log-P -4 to 9 (~95% of compounds).
• Orthogonal selectivity is missing, leaving highly polar and extremely hydrophobic regions underrepresented.

3

Analysis of Chemical Coverage

PCA and K-Means clustering results

---

Photo by Deng Xiang on Unsplash

Complete Chemical Coverage Remains Challenging

Current LC workflows provide substantial but limited coverage; comprehensive characterization requires integrating additional analytical modalities (GC, SFC, IC).

---

Photo by Nastuh Abootalebi on Unsplash

• PCA reveals high similarity in the majority of LC methods, dominated by RPLC setups.
• K-Means clustering (k=5) separates methods primarily by selectivity (RPLC vs HILIC) and eluent modifiers.
• Chemical space coverage: Methods span a wide theoretical domain (TPSA 0-852 Ų, Log-P -11 to 26).
• Adequate coverage: Most methods (>94%) are restricted to TPSA 0-200 Ų and Log-P -4 to 9.
• Comparison: The CompTox dataset (800k chemicals) highlights that extreme polar and hydrophobic regions remain largely inaccessible with current setups.

3

Chemical Coverage Analysis

Visualizing methodological similarity and chemical space coverage.

---

Photo by Tom Parkes on Unsplash

Comprehensive chemical space coverage requires orthogonal selectivity and wider methodological scope.

Advancing Exposome Monitoring

---

Photo by Jakub Żerdzicki on Unsplash

• PCA revealed substantial methodological similarity across the 236 LC setups.
• Primary variance driven by the dominant use of RPLC (89%) over HILIC (11%).
• K-means clustering (k=5) confirmed limited diversity in current chromatographic methods.
• Most methods are highly similar, differing mainly by particle size, temperature, and eluent modifiers.

• 0-852 Ų: TPSA Range
• -11 to 26.6: Log-P Range
• ~95%: Adequate Coverage

• LC-HRMS workflows are effective but fall short of comprehensive exposome coverage.
• Future work: Focus on purely NTA-generated data and more diverse chromatographic modes (e.g., GC, SFC, IC).
• Improved method reporting: Including pH, detailed gradients, and column parameters is crucial for future data-driven analyses.

> Current LC workflows, although effective within a broad region, remain far from achieving comprehensive characterization of chemical space.

— Jens Heemskerk (Thesis Conclusions)

---

Photo by Logan Voss on Unsplash

4

Conclusions & Future Perspectives

Final assessment and roadmap for future exposome research.

---

Photo by Tom Parkes on Unsplash

Questions?

Thank you for your attention

---

Photo by Trnava University on Unsplash

• Current LC workflows are effective but limited in comprehensive coverage.
• Lack of orthogonal selectivity (HILIC/IC) restricts measurements of extreme polar/hydrophobic regions.
• Complete chemical space coverage is not realistic under current conditions.
• Future: Focus on NTA-generated datasets and diversify chromatographic platforms (GC, SFC, IC).

Thesis Work Completed: Dec 2025

Thank you for your attention.

---

Photo by Nastuh Abootalebi on Unsplash