Performance evaluation of non-targeted peak-based cross-sample analysis for comprehensive two-dimensional gas chromatography-mass spectrometry data and application to processed hazelnut profiling

Johannes Kiefl, Chiara Cordero, Luca Nicolotti, Peter Schieberle, Stephen E Reichenbach, Carlo Bicchi

Research output: Contribution to journalArticle

30 Citations (Scopus)

Abstract

The continuous interest in non-targeted profiling induced the development of tools for automated cross-sample analysis. Such tools were found to be selective or not comprehensive thus delivering a biased view on the qualitative/quantitative peak distribution across 2D sample chromatograms. Therefore, the performance of non-targeted approaches needs to be critically evaluated. This study focused on the development of a validation procedure for non-targeted, peak-based, GC×GC-MS data profiling. The procedure introduced performance parameters such as specificity, precision, accuracy, and uncertainty for a profiling method known as Comprehensive Template Matching. The performance was assessed by applying a three-week validation protocol based on CITAC/EURACHEM guidelines. Optimized 1D and 2D retention times search windows, MS match factor threshold, detection threshold, and template threshold were evolved from two training sets by a semi-automated learning process. The effectiveness of proposed settings to consistently match 2D peak patterns was established by evaluating the rate of mismatched peaks and was expressed in terms of results accuracy. The study utilized 23 different 2D peak patterns providing the chemical fingerprints of raw and roasted hazelnuts (Corylus avellana L.) from different geographical origins, of diverse varieties and different roasting degrees. The validation results show that non-targeted peak-based profiling can be reliable with error rates lower than 10% independent of the degree of analytical variance. The optimized Comprehensive Template Matching procedure was employed to study hazelnut roasting profiles and in particular to find marker compounds strongly dependent on the thermal treatment, and to establish the correlation of potential marker compounds to geographical origin and variety/cultivar and finally to reveal the characteristic release of aroma active compounds.

Original languageEnglish (US)
Pages (from-to)81-90
Number of pages10
JournalJournal of Chromatography A
Volume1243
DOIs
StatePublished - Jun 22 2012

Fingerprint

Corylus
Template matching
Gas chromatography
Gas Chromatography-Mass Spectrometry
Mass spectrometry
Heat treatment
Dermatoglyphics
Uncertainty
Hot Temperature
Learning
Guidelines

Keywords

  • Comprehensive Template Matching fingerprinting
  • Corylus avellana L. volatile fraction
  • GC×GC-MS
  • Key aroma markers
  • Roasted hazelnut
  • Validation

ASJC Scopus subject areas

  • Analytical Chemistry
  • Biochemistry
  • Organic Chemistry

Cite this

Performance evaluation of non-targeted peak-based cross-sample analysis for comprehensive two-dimensional gas chromatography-mass spectrometry data and application to processed hazelnut profiling. / Kiefl, Johannes; Cordero, Chiara; Nicolotti, Luca; Schieberle, Peter; Reichenbach, Stephen E; Bicchi, Carlo.

In: Journal of Chromatography A, Vol. 1243, 22.06.2012, p. 81-90.

Research output: Contribution to journalArticle

@article{b77053461ed248f2821086ef8f29e690,
title = "Performance evaluation of non-targeted peak-based cross-sample analysis for comprehensive two-dimensional gas chromatography-mass spectrometry data and application to processed hazelnut profiling",
abstract = "The continuous interest in non-targeted profiling induced the development of tools for automated cross-sample analysis. Such tools were found to be selective or not comprehensive thus delivering a biased view on the qualitative/quantitative peak distribution across 2D sample chromatograms. Therefore, the performance of non-targeted approaches needs to be critically evaluated. This study focused on the development of a validation procedure for non-targeted, peak-based, GC×GC-MS data profiling. The procedure introduced performance parameters such as specificity, precision, accuracy, and uncertainty for a profiling method known as Comprehensive Template Matching. The performance was assessed by applying a three-week validation protocol based on CITAC/EURACHEM guidelines. Optimized 1D and 2D retention times search windows, MS match factor threshold, detection threshold, and template threshold were evolved from two training sets by a semi-automated learning process. The effectiveness of proposed settings to consistently match 2D peak patterns was established by evaluating the rate of mismatched peaks and was expressed in terms of results accuracy. The study utilized 23 different 2D peak patterns providing the chemical fingerprints of raw and roasted hazelnuts (Corylus avellana L.) from different geographical origins, of diverse varieties and different roasting degrees. The validation results show that non-targeted peak-based profiling can be reliable with error rates lower than 10{\%} independent of the degree of analytical variance. The optimized Comprehensive Template Matching procedure was employed to study hazelnut roasting profiles and in particular to find marker compounds strongly dependent on the thermal treatment, and to establish the correlation of potential marker compounds to geographical origin and variety/cultivar and finally to reveal the characteristic release of aroma active compounds.",
keywords = "Comprehensive Template Matching fingerprinting, Corylus avellana L. volatile fraction, GC×GC-MS, Key aroma markers, Roasted hazelnut, Validation",
author = "Johannes Kiefl and Chiara Cordero and Luca Nicolotti and Peter Schieberle and Reichenbach, {Stephen E} and Carlo Bicchi",
year = "2012",
month = "6",
day = "22",
doi = "10.1016/j.chroma.2012.04.048",
language = "English (US)",
volume = "1243",
pages = "81--90",
journal = "Journal of Chromatography",
issn = "0021-9673",
publisher = "Elsevier",

}

TY - JOUR

T1 - Performance evaluation of non-targeted peak-based cross-sample analysis for comprehensive two-dimensional gas chromatography-mass spectrometry data and application to processed hazelnut profiling

AU - Kiefl, Johannes

AU - Cordero, Chiara

AU - Nicolotti, Luca

AU - Schieberle, Peter

AU - Reichenbach, Stephen E

AU - Bicchi, Carlo

PY - 2012/6/22

Y1 - 2012/6/22

N2 - The continuous interest in non-targeted profiling induced the development of tools for automated cross-sample analysis. Such tools were found to be selective or not comprehensive thus delivering a biased view on the qualitative/quantitative peak distribution across 2D sample chromatograms. Therefore, the performance of non-targeted approaches needs to be critically evaluated. This study focused on the development of a validation procedure for non-targeted, peak-based, GC×GC-MS data profiling. The procedure introduced performance parameters such as specificity, precision, accuracy, and uncertainty for a profiling method known as Comprehensive Template Matching. The performance was assessed by applying a three-week validation protocol based on CITAC/EURACHEM guidelines. Optimized 1D and 2D retention times search windows, MS match factor threshold, detection threshold, and template threshold were evolved from two training sets by a semi-automated learning process. The effectiveness of proposed settings to consistently match 2D peak patterns was established by evaluating the rate of mismatched peaks and was expressed in terms of results accuracy. The study utilized 23 different 2D peak patterns providing the chemical fingerprints of raw and roasted hazelnuts (Corylus avellana L.) from different geographical origins, of diverse varieties and different roasting degrees. The validation results show that non-targeted peak-based profiling can be reliable with error rates lower than 10% independent of the degree of analytical variance. The optimized Comprehensive Template Matching procedure was employed to study hazelnut roasting profiles and in particular to find marker compounds strongly dependent on the thermal treatment, and to establish the correlation of potential marker compounds to geographical origin and variety/cultivar and finally to reveal the characteristic release of aroma active compounds.

AB - The continuous interest in non-targeted profiling induced the development of tools for automated cross-sample analysis. Such tools were found to be selective or not comprehensive thus delivering a biased view on the qualitative/quantitative peak distribution across 2D sample chromatograms. Therefore, the performance of non-targeted approaches needs to be critically evaluated. This study focused on the development of a validation procedure for non-targeted, peak-based, GC×GC-MS data profiling. The procedure introduced performance parameters such as specificity, precision, accuracy, and uncertainty for a profiling method known as Comprehensive Template Matching. The performance was assessed by applying a three-week validation protocol based on CITAC/EURACHEM guidelines. Optimized 1D and 2D retention times search windows, MS match factor threshold, detection threshold, and template threshold were evolved from two training sets by a semi-automated learning process. The effectiveness of proposed settings to consistently match 2D peak patterns was established by evaluating the rate of mismatched peaks and was expressed in terms of results accuracy. The study utilized 23 different 2D peak patterns providing the chemical fingerprints of raw and roasted hazelnuts (Corylus avellana L.) from different geographical origins, of diverse varieties and different roasting degrees. The validation results show that non-targeted peak-based profiling can be reliable with error rates lower than 10% independent of the degree of analytical variance. The optimized Comprehensive Template Matching procedure was employed to study hazelnut roasting profiles and in particular to find marker compounds strongly dependent on the thermal treatment, and to establish the correlation of potential marker compounds to geographical origin and variety/cultivar and finally to reveal the characteristic release of aroma active compounds.

KW - Comprehensive Template Matching fingerprinting

KW - Corylus avellana L. volatile fraction

KW - GC×GC-MS

KW - Key aroma markers

KW - Roasted hazelnut

KW - Validation

UR - http://www.scopus.com/inward/record.url?scp=84861462057&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84861462057&partnerID=8YFLogxK

U2 - 10.1016/j.chroma.2012.04.048

DO - 10.1016/j.chroma.2012.04.048

M3 - Article

VL - 1243

SP - 81

EP - 90

JO - Journal of Chromatography

JF - Journal of Chromatography

SN - 0021-9673

ER -