Multivariate Data Analysis in the Interpretation of GC-MS Data of Vegetable Oils and Animal Fats

doi:10.13560/j.cnki.biotech.bull.1985.2024-0525

Abstract

Abstract:

【Objective】This study is aimed to develop a comprehensive approach based on gas chromatography-mass spectrometry（GC-MS）combined with multivariate resolution and multivariate data analysis to achieve rapid and accurate identification of commonly encountered aged vegetable oils and animal fats in forensic science, particularly for samples with complex fatty acid compositions post-degradation that are difficult to distinguish through traditional spectral comparison.【Method】Firstly, the heuristic evolving latent projection（HELP）method was employed to resolve complex overlapping peaks acquired by GC-MS, enabling the separation and extraction of pure chromatograms and mass spectra of individual chemical components in vegetable oils and animal fats. Subsequently, two unsupervised learning methods, hierarchical cluster analysis（HCA）and principal component analysis（PCA）, were applied to reduce the dimensionality and perform cluster analysis of GC-MS data from 13 different types of aged vegetable oils and animal fats（aged at 60℃ for 36 d）attached to five different carriers, with the aim of exploring differences among species. Furthermore, orthogonal partial least squares-discriminant analysis（OPLS-DA）, a supervised learning method, was utilized for rapid identification of the geographical origin and brand of the fat and oil samples.【Result】The analyzed results via HCA and PCA indicated that this approach effectively differentiated the species categories of aged vegetable oils and animal fats. However, limitations were observed in further distinguishing fats and oils from different regions or brands. In contrast, the OPLS-DA model demonstrated higher classification accuracy, successfully achieving rapid and accurate identification of aged vegetable oils and animal fats from various regions or brands.【Conclusion】This study provides an efficient and accurate technical solution for the identification of aged vegetable oils and animal fats in forensic science through the integration of GC-MS with HELP multivariate resolution techniques, as well as HCA, PCA, and OPLS-DA analytical methods. This approach effectively addresses the complexities associated with oil degradation and decay, enabling rapid and precise differentiation of oils from different regions or brands.

Key words: multivariate curve resolution, multivariate data analysis, GC-MS, vegetable oils and animal fats, discrimination

LIU Ping-yang, LIU Zhan-fang, ZHOU Hong, ZHANG Guan-nan, SUN Zhen-wen, LI Ya-jun, ZHOU Zheng, LIU Yao. Multivariate Data Analysis in the Interpretation of GC-MS Data of Vegetable Oils and Animal Fats[J]. Biotechnology Bulletin, 2025, 41(1): 120-131.

Figures/Tables 7

Table 1 Vegetable oils and animal fats samples collected in this study

种类Category	产地/品牌 Origin/Brand	样本数量 Samples	分组 Group	种类 Category	产地/品牌 Origin/Brand	样本数量 Samples	分组 Group
Soybean oil	Inner Mongolia	8	A	Tung oil	Henan Province	8	G
	Heilongjiang Province	8			Hubei Province	8
	Anhui Province	8			Shanghai City	8
	Shandong Province	8		Castor oil	Hebei Province	8	H
	Sichuan Province	8			India/ Moutain rose herbs	8
	Guangdong Province	8			Egypt/ Ruifantine	8
	Shanxi Province	8			Hebei Province/Yanwei	8
	Yunnan Province	8			Henan Province/Huangjia	8
	Jilin Province	8			Guangdong Province	8
	Hubei Province	8		Butter	Henan Province	8	I
Olive oil	Spain/Oliveira	8	B		Guangdong Province	8
	Italy/Philippe Berry	8			Australia	8
	Spain/Ajinomoto	8			Inner Mongolia	8
	Spain/Obera	8			Shandong Province	8
	Spain/Earl	8			Beijing City	8
	Spain/Omanti	8			Sichuan Province	8
	Spain/Borges	8		Sheep oil	Shanghai City	8	J
Sunflower oil	Beijing City	8	C		Gansu Province	8
	Ukraine/HADAY	8			Xinjiang Uygur Autonomous Region	8
	Europe/COFCO	8			Jiangsu Province	8
	Europe/Arawana	8			Ningxia Hui Autonomous Region	8
	Ukraine/Dianxue	8			Shandong Province	8
	Russia/Red	8			Inner Mongolia	8
	Ukraine/Qiandaoyuan	8			Beijing City	8
Peanut oil	Shandong Province/Luhua	8	D		Shandong Province	8
	Inner Mongolia	8		Chicken oil	Yongjia/Sichuan Province	8	K
	Shandong Province	8		Chicken oil	Jinluo/Shandong Province	8	K
	Yunnan Province	8		Duck oil	Beijing City	8	L
	Guangdong Province	8		Lard	Inner Mongolia	8	M
	Beijing City	8			Xizang	8
Palm oil	Guangdong Province	8	E		Heilongjiang Province	8
	Indonesia	8			Hubei Province	8
	Brazil/ TMEPEREX	8			Guangdong Province	8
	Thailand/Huarui	8			Beijing City	8
	Malaysia/Tianyijia	8			Guizhou Province	8
Linseed oil	Inner Mongolia	8	F		Sichuan Province	8
	Ningxia Hui Autonomous Region/Haoyu	8			Gansu Province	8
	Ningxia Hui Autonomous Region/Yixiayuan	8			Yunnan Province	8
	China/Yuedushoufang	8
	Russia/Fuyide	8
	China/Qiumanxian	8

Table 1 Vegetable oils and animal fats samples collected in this study

种类Category	产地/品牌 Origin/Brand	样本数量 Samples	分组 Group	种类 Category	产地/品牌 Origin/Brand	样本数量 Samples	分组 Group
Soybean oil	Inner Mongolia	8	A	Tung oil	Henan Province	8	G
	Heilongjiang Province	8			Hubei Province	8
	Anhui Province	8			Shanghai City	8
	Shandong Province	8		Castor oil	Hebei Province	8	H
	Sichuan Province	8			India/ Moutain rose herbs	8
	Guangdong Province	8			Egypt/ Ruifantine	8
	Shanxi Province	8			Hebei Province/Yanwei	8
	Yunnan Province	8			Henan Province/Huangjia	8
	Jilin Province	8			Guangdong Province	8
	Hubei Province	8		Butter	Henan Province	8	I
Olive oil	Spain/Oliveira	8	B		Guangdong Province	8
	Italy/Philippe Berry	8			Australia	8
	Spain/Ajinomoto	8			Inner Mongolia	8
	Spain/Obera	8			Shandong Province	8
	Spain/Earl	8			Beijing City	8
	Spain/Omanti	8			Sichuan Province	8
	Spain/Borges	8		Sheep oil	Shanghai City	8	J
Sunflower oil	Beijing City	8	C		Gansu Province	8
	Ukraine/HADAY	8			Xinjiang Uygur Autonomous Region	8
	Europe/COFCO	8			Jiangsu Province	8
	Europe/Arawana	8			Ningxia Hui Autonomous Region	8
	Ukraine/Dianxue	8			Shandong Province	8
	Russia/Red	8			Inner Mongolia	8
	Ukraine/Qiandaoyuan	8			Beijing City	8
Peanut oil	Shandong Province/Luhua	8	D		Shandong Province	8
	Inner Mongolia	8		Chicken oil	Yongjia/Sichuan Province	8	K
	Shandong Province	8		Chicken oil	Jinluo/Shandong Province	8	K
	Yunnan Province	8		Duck oil	Beijing City	8	L
	Guangdong Province	8		Lard	Inner Mongolia	8	M
	Beijing City	8			Xizang	8
Palm oil	Guangdong Province	8	E		Heilongjiang Province	8
	Indonesia	8			Hubei Province	8
	Brazil/ TMEPEREX	8			Guangdong Province	8
	Thailand/Huarui	8			Beijing City	8
	Malaysia/Tianyijia	8			Guizhou Province	8
Linseed oil	Inner Mongolia	8	F		Sichuan Province	8
	Ningxia Hui Autonomous Region/Haoyu	8			Gansu Province	8
	Ningxia Hui Autonomous Region/Yixiayuan	8			Yunnan Province	8
	China/Yuedushoufang	8
	Russia/Fuyide	8
	China/Qiumanxian	8

Fig. 1 Analysis of the chromatographic and mass spectrometric characteristics of vegetable oils and animal fats and their regional differences A: TIC of 13 different types of vegetable oils and animal fats before aging. B: Mass spectrometry of methyl palmitate. C: Mass spectrometry of methyl stearate. D: Mass spectrometry of methyl oleate. E: Mass spectrometry of methyl linoleate. F: Comparison of chromatograms of sheep oil from Jiangsu Province, Xinjiang Uygur Autonomous Region, and Shandong Province before aging

Fig. 2 GC-MS analysis of fatty acid composition and its changes of olive oil under different aging conditions A: Changes of fatty acid composition of olive oil at 0, 6, 12, 18, 24, 30, and 36 d in the incubator of 60℃. B: The trend of changes of saturated fatty acids and unsaturated fatty acids of olive oil at 0, 6, 12, 18, 24, 30, and 36 d in the incubator of 60℃. C: GC-MS chromatogram of fresh olive oil. D-F: GC-MS chromatogram of aged olive oil placed on the experimental bench after 6, 18, and 36 d of indoor air exposure. G-I: GC-MS chromatogram of aged olive oil placed in the incubator of 60℃ after 6, 18, and 36 d. The carrier includes A4 paper, corrugated paper, sponge, soil, and floor leather, and the data are taken as their average values（n=3）

Fig. 3 Schematic diagram of data imbalance handling principles A: Schematic diagram of SMOTE generated samples. B: Schematic diagram of Tomek links. C: Schematic diagram of SMOTE-Tomek links

Fig. 4 HCA of vegetable oils and animal fats a: HCA of 13 different types of vegetable oils and animal fats. b: HCA of 5 different types of animal fats. c: HCA of 8 different types of vegetable oils. The carrier includes A4 paper, corrugated paper, sponge, soil, and floor leather, and the data is taken as the average value（n=3）. A: Soybean oil. B: Olive oil. C: Sunflower oil. D: Peanut oil. E: Palm oil. F: Linseed oil. G: Tung oil. H: Castor oil. I: Butter. J: Sheep oil. K: Chicken oil. L: Duck oil. M: Lard

Fig. 5 HELP analysis of vegetable oils and animal fats and their principal component analysis A: GC-MS total ion flow chromatogram and local magnification of aged（60℃ 36 d）tung oil. B: Total ion flow chromatogram of aged tung oil between 24.967-25.127 min. C: Two-dimensional data graph of aged tung oil between 24.967-25.127 min. D: Rank plot of aged tung oil between 24.967-25.127 min[Lines below the dotted line indicate noise level, and lines above dotted line indicate the local ranks（Number of compounds）]. E: The chromatogram of pure compound was analyzed by the HELP method. F: PCA score plot of 13 vegetable oils and animal fats aged on 5 carriers（sponge, A4 printer paper, floor leather, soil, and corrugated paper）, with R2X（cum）=0.859 and Q2（cum）=0.815. G: The corresponding PCA load diagram. The carrier includes A4 paper, corrugated paper, sponge, soil, and floor leather, and the data are taken as their average values（n=3）

Fig. 6 PCA and OPLS-DA of aged sheep oil（60℃ 36 d）in 8 provinces A: PCA model of aged sheep oil（60℃ 36 d）in 8 provinces（Gansu, Inner Mongolia, Ningxia, Jiangsu, Xinjiang, Beijing, Shanghai, and Shandong）. B: OPLS-DA model of aged sheep oil（60℃ 36 d）in 8 provinces（Gansu, Inner Mongolia, Ningxia, Jiangsu, Xinjiang, Beijing, Shanghai, and Shandong）. C: Permutation test result of the OPLS-DA model. D: Model parameter of OPLS-DA. The carrier includes A4 paper, corrugated paper, sponge, soil, and floor leather, and the data are taken as their average values（n=3）

References 36

[1]	吴成招, 王一韬, 胡栋, 等. 近红外光谱技术在食用油掺伪检测中的研究进展[J]. 光谱学与光谱分析, 2023, 43(3): 685-691.
	Wu CZ, Wang YT, Hu D, et al. Research progress of near-infrared spectroscopy in the detection of edible oil adulteration[J]. Spectrosc Spectr Anal, 2023, 43(3): 685-691.
[2]	Tan CH, Kong I, Irfan U, et al. Edible oils adulteration: a review on regulatory compliance and its detection technologies[J]. J Oleo Sci, 2021, 70(10): 1343-1356. doi: 10.5650/jos.ess21109 pmid: 34497179
[3]	中华人民共和国国家卫生和计划生育委员会, 国家食品药品监督管理总局. 食品中脂肪酸的测定: GB 5009.168—2016[S]. 北京: 中国标准出版社, 2016.
	National Health and Family Planning Commission of the People's Republic of China, China Food and Drug Administration. Determination of Fatty Acids in Foods:GB 5009.168-2016[S]. Beijing: Standards Press of China, 2016.
[4]	ASTM International. Standard test method for extraction and derivatization of vegetable oils and fats from fire debris and liquid samples with analysis by gas chromatography-mass spectrometry: E 2881-18[S]. West Conshohocken, PA: ASTM International, 2018.
[5]	Sousa Queirós S, von der Lühe B, Silva-Bessa A, et al. Lipidic compounds found in soils surrounding human decomposing bodies and its use in forensic investigations - A narrative review[J]. Sci Justice, 2023, 63(3): 303-312. doi: 10.1016/j.scijus.2023.02.001 pmid: 37169455
[6]	Hussain MN, Basri KN, Arshad S, et al. Analysis of lard in palm oil using long-wave near-infrared(LW-NIR)spectroscopy and gas chromatography-mass spectroscopy(GC-MS)[J]. Food Anal Meth, 2023, 16(2): 349-355.
[7]	Rebiai A, Hemmami H, Zeghoud S, et al. Current application of chemometrics analysis in authentication of natural products: a review[J]. Comb Chem High Throughput Screen, 2022, 25(6): 945-972.
[8]	Balbino S, Vincek D, Trtanj I, et al. Assessment of pumpkin seed oil adulteration supported by multivariate analysis: comparison of GC-MS, colourimetry and NIR spectroscopy data[J]. Foods, 2022, 11(6): 835.
[9]	Üçüncüoğlu D, Küçük VA. A multivariate data approach for FTIR-ATR monitoring of virgin olive oils: Lipid structure during autoxidation[J]. OCL, 2019, 26: 42.
[10]	范璐, 吴娜娜, 霍权恭, 等. 模式识别法分析5种植物油脂[J]. 分析化学, 2008, 36(8): 1133-1137.
	Fan L, Wu NN, Huo QG, et al. Pattern recognition analysis of 5 kinds of vegetable oil and fat by the content of fatty acids[J]. Chin J Anal Chem, 2008, 36(8): 1133-1137.
[11]	马路凯, 崔芸, 刘漩, 等. 植物油中不饱和醛作为特征氧化指标的适用性[J]. 中国食品学报, 2022, 22(5): 239-248.
	Ma LK, Cui Y, Liu X, et al. Applicability of unsaturated aldehyde in vegetable oil as characteristic oxidation index[J]. J Chin Inst Food Sci Technol, 2022, 22(5): 239-248.
[12]	刘平阳, 刘占芳, 周红, 等. 基于气相色谱-质谱的脂质组学方法鉴别动植物油脂[J/OL]. 分析试验室, 2024. http://kns.cnki.net/kcms/detail/11.2017.TF.20240521.1042.008.html.
	Liu PY, Liu ZF, Zhou H, et al. Gas chromatography-mass spectrometrybased lipidomics approach for discrimination of vegetable oils and animal fats[J/OL]. Chin J Anal Lab, 2024. http://kns.cnki.net/kcms/detail/11.2017.TF.20240521.1042.008.html.
[13]	时秋娜, 刘占芳, 朱军, 等. 全二维气相色谱-质谱法检测动植物油[J]. 中国油脂, 2017, 42(6): 138-142.
	Shi QN, Liu ZF, Zhu J, et al. Detection of fatty acids in animal fats and vegetable oils by comprehensive two-dimensional gas chromatography coupled with mass spectrometry[J]. China Oils Fats, 2017, 42(6): 138-142.
[14]	Hantao LW, Aleme HG, Pedroso MP, et al. Multivariate curve resolution combined with gas chromatography to enhance analytical separation in complex samples: a review[J]. Anal Chim Acta, 2012, 731: 11-23. doi: 10.1016/j.aca.2012.04.003 pmid: 22652260
[15]	Xin ZQ, Ren DB, Zhang XJ, et al. Chromatographic fingerprints combined with chemometric methods reveal the chemical features of authentic radix polygalae[J]. J AOAC Int, 2017, 100(1): 30-37. doi: 10.5740/jaoacint.16-0225 pmid: 27743444
[16]	欧高炎, 朱占星, 董彬, 等. 数据科学导引[M]. 北京: 高等教育出版社, 2017: 29-45.
	Ou GY, Zhu ZX, Dong B, et al. Introduction to data science[M]. Beijing: Higher Education Press, 2017: 29-45.
[17]	时秋娜. 常见动植物油脂及人体脂肪的GC-MS法分析[J]. 环境化学, 2023, 42(10): 3600-3602.
	Shi QN. Analysis of common animal and plant oils and human fat by GC-MS[J]. Environ Chem, 2023, 42(10): 3600-3602.
[18]	Richaud E, Audouin L, Fayolle B, et al. Rate constants of oxidation of unsaturated fatty esters studied by chemiluminescence[J]. Chem Phys Lipids, 2012, 165(7): 753-759. doi: 10.1016/j.chemphyslip.2012.09.002 pmid: 23026153
[19]	Ding C, Wang L, Yao YP, et al. Mechanism of the initial oxidation of monounsaturated fatty acids[J]. Food Chem, 2022, 392: 133298.
[20]	Jeong DH, Kim SE, Choi WH, et al. A comparative study on the influence of undersampling and oversampling techniques for the classification of physical activities using an imbalanced accelerometer dataset[J]. Healthcare, 2022, 10(7): 1255.
[21]	García S, Herrera F. Evolutionary undersampling for classification with imbalanced datasets: proposals and taxonomy[J]. Evol Comput, 2009, 17(3): 275-306. doi: 10.1162/evco.2009.17.3.275 pmid: 19708770
[22]	Yi XK, Xu YY, Hu Q, et al. ASN-SMOTE: a synthetic minority oversampling method with adaptive qualified synthesizer selection[J]. Complex Intell Syst, 2022, 8(3): 2247-2272.
[23]	Ning Q, Zhao XW, Ma ZQ. A novel method for identification of glutarylation sites combining borderline-SMOTE with tomek links technique in imbalanced data[J]. IEEE/ACM Trans Comput Biol Bioinform, 2022, 19(5): 2632-2641.
[24]	Sasada T, Liu ZY, Baba T, et al. A resampling method for imbalanced datasets considering noise and overlap[J]. Procedia Comput Sci, 2020, 176: 420-429.
[25]	Todeschini R, Ballabio D, Termopoli V, et al. Extended multivariate comparison of 68 cluster validity indices. A review[J]. Chemom Intell Lab Syst, 2024, 251: 105117.
[26]	Wang FS, Chen JG, Liu FR. Keyframe generation method via improved clustering and silhouette coefficient for video summarization[J]. J Web Eng, 2021, 20(1): 147-170.
[27]	Robert V, Vasseur Y, Brault V. Comparing high-dimensional partitions with the co-clustering adjusted rand index[J]. J Classif, 2021, 38(1): 158-186.
[28]	Zielinski AAF, Haminiuk CWI, Nunes CA, et al. Chemical composition, sensory properties, provenance, and bioactivity of fruit juices as assessed by chemometrics: a critical review and guideline[J]. Compr Rev Food Sci Food Saf, 2014, 13(3): 300-316.
[29]	Jafarzadegan M, Safi-Esfahani F, Beheshti Z. Combining hierarchical clustering approaches using the PCA method[J]. Expert Syst Appl, 2019, 137: 1-10. doi: 10.1016/j.eswa.2019.06.064
[30]	Chen WH, Xu YH, Yu ZW, et al. Hybrid dimensionality reduction forest with pruning for high-dimensional data classification[J]. IEEE Access, 2020, 8: 40138-40150.
[31]	Wang ZY, Han BS, Jing WF, et al. Effects of different steeping temperatures on the leaching of aroma components in black tea by SPME-GC-MS coupled with chemometric method[J]. J AOAC Int, 2019, 102(6): 1834-1844. doi: 10.5740/jaoacint.18-0405 pmid: 30795823
[32]	He M, Yang ZY, Yang TB, et al. Chemometrics-enhanced one-dimensional/comprehensive two-dimensional gas chromatographic analysis for bioactive terpenoids and phthalides in Chaihu Shugan San essential oils[J]. J Chromatogr B Analyt Technol Biomed Life Sci, 2017, 1052: 158-168. doi: S1570-0232(16)31494-5 pmid: 28390283
[33]	Amalia L, Yuliana ND, Sugita P, et al. Volatile compounds, texture, and color characterization of meatballs made from beef, rat, wild boar, and their mixtures[J]. Heliyon, 2022, 8(10): e10882.
[34]	Deng JK, Zhao HB, Qi B, et al. Volatile characterization of crude and refined walnut oils from aqueous enzymatic extraction by GC-IMS and GC-MS[J]. Arab J Chem, 2024, 17(1): 105404.
[35]	Worley B, Powers R. PCA as a practical indicator of OPLS-DA model reliability[J]. Curr Metabolomics, 2016, 4(2): 97-103.
[36]	Triba MN, Le Moyec L, Amathieu R, et al. PLS/OPLS models in metabolomics: the impact of permutation of dataset rows on the K-fold cross-validation quality parameters[J]. Mol Biosyst, 2015, 11(1): 13-19. doi: 10.1039/c4mb00414k pmid: 25382277