Metabolomic Insight into Soy Sauce using H NMR Spectroscopy
Page S-1
1
1
2
1
1
Bong-Kuk Ko, Hyuk-Jin Ahn, Frans van den Berg, Cherl-Ho Lee, and Young-Shick Hong 1
School of Life Science and Biotechnology, Korea University, 5-1, Anam-dong, Sungbuk-gu, Seoul 1362
701, Republic of Korea, and Department of Food Sciences, Faculty of Life Sciences, University of Copenhagen, Denmark
Supporting Information for:
Metabolomic Insight into Soy Sauce using 1H NMR Spectroscopy
Figure of Contents:
Figure S-1. An Illustration of traditional Korean soy sauce production. (A) Natural inoculation of various microorganisms from the natural environment and rice straw into meju, as a starter for fermentation, for 50 days; (B) Meju with molds and bacteria; (C) Brine fermentation of meju in onggi tank, Korean ceramic earthenware jars for two months; (D) Aging of soy sauce in onggi for at least 1 year, following separation of soybean paste and soy sauce; (E) End product of soy sauce to be commercialized. (F) reveals each process step in producing traditional Korean soy sauce (page S-3).
Figure S-2. Two-dimensional (2D) 1H-1H TOCSY NMR spectrum of traditional Korean soy sauce aged for 12 years. The 2D contour plot (B) corresponds to 1D spectrum acquired using NOESYPRESAT pulse sequence (A). Solid rectangles in panel B trace the 1H-covalent network of protons as pairs of symmetric cross-peaks with respect to the diagonal axis. Most metabolites were also assigned by spiking experiments. Key: 1, leucine; 2, isoleucine; 3, valine; 4, lactate; 5, alanine; 6, acetate; 7, arginine; 8, betaine; 9, tyramine; 10, phenylalanine; 11, formate; 14, lysine; 15, γ-aminobutyrate (GABA); 17, ethanol; 18, succinate; 19, tyrosine; 20, proline; 21, choline; 22, uracil; 23, aspartate; 24, methionine; 25, pyroglutamate; 26, glutamate; 27, glycine; 28, trimethylamine; 29, malonate; 30, phosphocholine; 32, oligosaccharide (O); 33, oligosaccharide (O); 34, oligosaccharide (O); 35, glycerol; 36, propionate; 37, hypoxanthine; U1, U2, U3, U4 and U5, unknown (page S-4).
S-1
Figure S-3. PCA score plots after normalizing 1H NMR spectra of soy sauces to total spectral area (A) and after normalization through division of each NMR spectrum by median spectrum (B). Black boxes and red circles represent commercial soy sauces from CJ and Sempio manufacturers. Green diamonds represent traditional Korean soy sauces aged for 1 year. Filled and open symbols were normalized in the absence and presence of ethanol peaks in the NMR spectra, respectively, demonstrating that ethanol with huge spectral intensity affects the normalization to total spectra area. PCA models were generated after removing ethanol peaks, following normalization (page S-5).
Figure S-4. PCA scores plots derived from traditional Korean soy sauces. Unboiled (U) soy sauces were aged for 1 year whereas other soy sauces were aged for 1, 2 and 4 years following boiling for 30 minutes. Scattered plots of unboiled soy sauces reveal more intra-variations than those of boiled soy sauces, resulting in their uncontrolled or inconstant quality (page S-6).
Figure S-5. PCA scores (A and C) and loadings (B and D) plots derived from 1H NMR spectra of traditional Korean soy sauces aged for 4 years (4 years), and commercial Korean (CJ) and Japanese (Kikkoman) soy sauces (page S-7).
S-2
Manufacture of Korea Traditional Sauce Sauces. Traditional Korean soy sauces were obtained from a cottage manufacturer who is nationally famous and produces soy sauces by traditional Korean methods. Their soy sauces have been passed down from generation to generation for two centuries. To briefly explain their manufacturing processes (Figure S-1 in the Supporting Information), dried soybeans were boiled and pasted into fine bits. This paste was then formed into blocks that were tied up with rice straw; these are called meju and serve as a starter for fermentation. The meju was then dried in the sunlight. After drying, the meju was moved to a warm place to speed up the fermentation and there it was fermented for 50 days. During this stage, various beneficial bacteria and molds from the rice straw and the natural environment transformed into the meju, and the meju then served as a traditional Korean soybean fermentation starter. The meju were put into large Korean ceramic earthenware jars, called onggi, which are traditional fermentation vessels in Korea (1), with brine and left to further ferment for 8 weeks. After fermentation in the brine, the liquid and meju were separated and the liquid became traditional Korean soy sauce, kanjang. The meju was subsequently crushed and became traditional homemade soybean paste, doenjang. After the separation and subsequent boiling, the traditional soy sauce was aged for 1, 2, 4 or 12 years that are used for this study.
Onggi Tank for fermentation of Korean Traditional Soy Sauce. Raw soy sauce is traditionally boiled for 30 minutes to stop growth of any undesirable microorganisms and to inactivate residual enzymes after filtration or pressing, resulting in stabilization in the flavor and color during storage (2). Recently, ceramic membranes have been used for microfiltration instead of boiling to provide commercial products with constant quality (3). Most cottage manufacturers, especially generational manufacturers in Korea, traditionally S-3
boil raw soy sauce to induce aging for at least 1 year and sometimes for over 10 years in an onggi tank (Figure S-1 in the Supporting Information), ensuring a unique and high-quality product. Onggi is made from ceramic material and is porous, which allows permeation of oxygen, carbon dioxide, water and salt during the fermentation (1). The porosity may provide useful growth conditions for microflora and cause them to be immobilized. The pore size of mesoporous molecular sieves also affects the activity of immobilized enzymes (4, 5). Immobilized cells in the pores of ceramic beads contribute to rapid ethanol fermentation for soy sauce (6). The main advantage of using immobilized cells in soy sauce was well reviewed (7). Recently, it has been reported that the porosity of onggi plays an important role in immobilizing microflora and enzymes and in providing high quality and high reproducibility when used for repeated-batch fermentation of soy sauce (1, 8, 9).
Normalization Effect of Soy Sauce NMR Spectra on PCA Models. In metabolomic studies, normalization is the preprocessing step for 1H NMR spectra that accounts for variations in the overall concentration of samples caused by different dilutions. Therefore, the normalization step is crucial to compensate for differences in the overall concentration. For the 1H NMR spectra of complex biofluids, such as urine and serum, integral normalization is the standard for normalizing NMR spectra (10, 11). The integral normalization procedure divides each signal of a spectrum by the integral of the complete spectrum. Another common procedure is creatinine normalization (12, 13), in which the NMR spectrum is divided by the peak area of the creatinine methyl resonance. In addition, each NMR spectrum can be divided by median spectrum for normalization (14, 15). In the present study, two methods were applied for normalizing the NMR spectra: normalization to total spectral area and normalization by dividing each spectrum by the median spectrum (Figure S-3). S-4
Figure S-1. An Illustration of traditional Korean soy sauce production. (A) Natural inoculation of various microorganisms from the natural environment and rice straw into meju, as a starter for fermentation, for 50 days; (B) Meju with molds and bacteria; (C) Brine fermentation of meju in onggi tank, Korean ceramic earthenware jars for two months; (D) Aging of soy sauce in onggi for at least 1 year, following separation of soybean paste and soy sauce; (E) End product of soy sauce to be commercialized. (F) reveals each process step in producing traditional Korean soy sauce.
S-5
Figure S-2. Two-dimensional (2D) 1H-1H TOCSY NMR spectrum of traditional Korean soy sauce aged for 12 years. The 2D contour plot (B) corresponds to 1D spectrum acquired using NOESYPRESAT pulse sequence (A). Solid rectangles in panel B trace the 1H-covalent network of protons as pairs of symmetric cross-peaks with respect to the diagonal axis. Most metabolites were also assigned by spiking experiments. Key: 1, leucine; 2, isoleucine; 3, valine; 4, lactate; 5, alanine; 6, acetate; 7, arginine; 8, betaine; 9, tyramine; 10, phenylalanine; 11, formate; 14, lysine; 15, γ-aminobutyrate (GABA); 17, ethanol; 18, succinate; 19, tyrosine; 20, proline; 21, choline; 22, uracil; 23, aspartate; 24, methionine; 25, pyroglutamate; 26, glutamate; 27, glycine; 28, trimethylamine; 29, malonate; 30, phosphocholine; 32, oligosaccharide (O1); 33, oligosaccharide (O2); 34, oligosaccharide (O3); 35, glycerol; 36, propionate; 37, hypoxanthine; U1, U2, U3, U4 and U5, unknown.
S-6
Figure S-3. PCA score plots after normalizing 1H NMR spectra of soy sauces to total spectral area (A) and after normalization through division of each NMR spectrum by median spectrum (B). Black boxes and red circles represent commercial soy sauces from CJ and Sempio manufacturers. Green diamonds represent traditional Korean soy sauces aged for 1 year. Filled and open symbols were normalized in the absence and presence of ethanol peaks in the NMR spectra, respectively, demonstrating that ethanol with huge spectral intensity affects the normalization to total spectra area. PCA models were generated after removing ethanol peaks, following normalization.
S-7
Figure S-4. PCA scores plots derived from traditional Korean soy sauces. Unboiled (U) soy sauces were aged for 1 year whereas other soy sauces were aged for 1, 2 and 4 years following boiling for 30 minutes. Scattered plots of unboiled soy sauces reveal more intra-variations than those of boiled soy sauces, resulting in their uncontrolled or inconstant quality.
S-8
Figure S-5. PCA scores (A and C) and loadings (B and D) plots derived from 1H NMR spectra of traditional Korean soy sauces aged for 4 years (4 years), and commercial Korean (CJ) and Japanese (Kikkoman) soy sauces.
S-9
LITERATURE CITED 1.
Lee, K. S.; Lee, Y. B.; Lee, D. S.; Chung, S. K. Quality evaluation of Korean soy sauce fermented in Korean earthenware (Onggi) with different glazes. Int. J. Food Sci. Technol. 2006, 41, 1158-1163.
2.
Luh, B. S. Industrial production of soy sauce. J. Ind. Microbiol. 1995, 14, 467-471.
3.
Li, M. S.; Zhao, Y. J.; Zhou, S. Y.; Xing, W. H.; Wong, F. S. Resistance analysis for ceramic membrane microfiltration of raw soy sauce. J. Membr. Sci. 2007, 299, 122-129.
4.
Yiu, H. H. P.; Wright, P. A.; Botting, N. P. Enzyme immobilisation using siliceous mesoporous molecular sieves. Micropor. Mesopor. Mat. 2001, 44, 763-768.
5.
Yiu, H. H. P.; Wright, P. A.; Botting, N. P. Enzyme immobilisation using SBA-15 mesoporous molecular sieves with functionalised surfaces. J. Mol. Catal. B-Enzym. 2001, 15, 81-92.
6.
Iwasaki, K.; Nakajima, M.; Sasahara, H.; Watanabe, A. Rapid ethanol fermentation for soy-sauce production by immobilized yeast-cells. Agric. Biol. Chem. 1991, 55, 22012207.
7.
Norton, S.; Vuillemard, J. C. Food bioconversions and metabolite production using immobilized cell technology. Crit. Rev. Biotechnol. 1994, 14, 193-224.
8.
Seo, G. H.; Yun, J. H.; Chung, S. K.; P., P. W.; Lee, D. S. Physical properties of Korean earthenware containers affected by soy sauce fermentation use. Food Sci. Biotechnol. 2006, 15, 168-172.
9.
Iwasaki, K. I.; Nakajima, M.; Sasahara, H. Porous alumina beads for immobilization of lactic-acid bacteria and its application for repeated-batch fermentation in soy-sauce production. J. Ferment. Bioeng. 1992, 73, 375-379.
S-10
10.
Bollard, M. E.; Stanley, E. G.; Lindon, J. C.; Nicholson, J. K.; Holmes, E. NMR-based metabonomic approaches for evaluating physiological influences on biofluid composition. NMR Biomed. 2005, 18, 143-162.
11.
Craig, A.; Cloareo, O.; Holmes, E.; Nicholson, J. K.; Lindon, J. C. Scaling and normalization effects in NMR spectroscopic metabonomic data sets. Anal. Chem. 2006, 78, 2262-2267.
12.
Prevot, A.; Martini, S.; Guignard, J. P. Rev. Med. Suisse Romande 2002, 122, 625-630.
13.
van Velzen, E. J. J.; Westerhuis, J. A.; van Duynhoven, J. P. M.; van Dorsten, F. A.; Hoefsloot, H. C. J.; Jacobs, D. M.; Smit, S.; Draijer, R.; Kroner, C. I.; Smilde, A. K. Multilevel data analysis of a crossover designed human nutritional intervention study. J. Proteome Res. 2008, 7, 4483-4491.
14.
Jansen, J. J.; Bro, R.; Hoefsloot, H. C. J.; van den Berg, F. W. J.; Westerhuis, J. A.; Smilde, A. K. PARAFASCA: ASCA combined with PARAFAC for the analysis of metabolic fingerprinting data. J. Chemometr. 2008, 22, 114-121.
15.
Dieterle, F.; Ross, A.; Schlotterbeck, G.; Senn, H. Probabilistic quotient normalization as robust method to account for dilution of complex biological mixtures. Application in H-1 NMR metabonomics. Anal. Chem. 2006, 78, 4281-4290.
S-11
1
1
2
1
1
Bong-Kuk Ko, Hyuk-Jin Ahn, Frans van den Berg, Cherl-Ho Lee, and Young-Shick Hong 1
School of Life Science and Biotechnology, Korea University, 5-1, Anam-dong, Sungbuk-gu, Seoul 1362
701, Republic of Korea, and Department of Food Sciences, Faculty of Life Sciences, University of Copenhagen, Denmark
Supporting Information for:
Metabolomic Insight into Soy Sauce using 1H NMR Spectroscopy
Figure of Contents:
Figure S-1. An Illustration of traditional Korean soy sauce production. (A) Natural inoculation of various microorganisms from the natural environment and rice straw into meju, as a starter for fermentation, for 50 days; (B) Meju with molds and bacteria; (C) Brine fermentation of meju in onggi tank, Korean ceramic earthenware jars for two months; (D) Aging of soy sauce in onggi for at least 1 year, following separation of soybean paste and soy sauce; (E) End product of soy sauce to be commercialized. (F) reveals each process step in producing traditional Korean soy sauce (page S-3).
Figure S-2. Two-dimensional (2D) 1H-1H TOCSY NMR spectrum of traditional Korean soy sauce aged for 12 years. The 2D contour plot (B) corresponds to 1D spectrum acquired using NOESYPRESAT pulse sequence (A). Solid rectangles in panel B trace the 1H-covalent network of protons as pairs of symmetric cross-peaks with respect to the diagonal axis. Most metabolites were also assigned by spiking experiments. Key: 1, leucine; 2, isoleucine; 3, valine; 4, lactate; 5, alanine; 6, acetate; 7, arginine; 8, betaine; 9, tyramine; 10, phenylalanine; 11, formate; 14, lysine; 15, γ-aminobutyrate (GABA); 17, ethanol; 18, succinate; 19, tyrosine; 20, proline; 21, choline; 22, uracil; 23, aspartate; 24, methionine; 25, pyroglutamate; 26, glutamate; 27, glycine; 28, trimethylamine; 29, malonate; 30, phosphocholine; 32, oligosaccharide (O); 33, oligosaccharide (O); 34, oligosaccharide (O); 35, glycerol; 36, propionate; 37, hypoxanthine; U1, U2, U3, U4 and U5, unknown (page S-4).
S-1
Figure S-3. PCA score plots after normalizing 1H NMR spectra of soy sauces to total spectral area (A) and after normalization through division of each NMR spectrum by median spectrum (B). Black boxes and red circles represent commercial soy sauces from CJ and Sempio manufacturers. Green diamonds represent traditional Korean soy sauces aged for 1 year. Filled and open symbols were normalized in the absence and presence of ethanol peaks in the NMR spectra, respectively, demonstrating that ethanol with huge spectral intensity affects the normalization to total spectra area. PCA models were generated after removing ethanol peaks, following normalization (page S-5).
Figure S-4. PCA scores plots derived from traditional Korean soy sauces. Unboiled (U) soy sauces were aged for 1 year whereas other soy sauces were aged for 1, 2 and 4 years following boiling for 30 minutes. Scattered plots of unboiled soy sauces reveal more intra-variations than those of boiled soy sauces, resulting in their uncontrolled or inconstant quality (page S-6).
Figure S-5. PCA scores (A and C) and loadings (B and D) plots derived from 1H NMR spectra of traditional Korean soy sauces aged for 4 years (4 years), and commercial Korean (CJ) and Japanese (Kikkoman) soy sauces (page S-7).
S-2
Manufacture of Korea Traditional Sauce Sauces. Traditional Korean soy sauces were obtained from a cottage manufacturer who is nationally famous and produces soy sauces by traditional Korean methods. Their soy sauces have been passed down from generation to generation for two centuries. To briefly explain their manufacturing processes (Figure S-1 in the Supporting Information), dried soybeans were boiled and pasted into fine bits. This paste was then formed into blocks that were tied up with rice straw; these are called meju and serve as a starter for fermentation. The meju was then dried in the sunlight. After drying, the meju was moved to a warm place to speed up the fermentation and there it was fermented for 50 days. During this stage, various beneficial bacteria and molds from the rice straw and the natural environment transformed into the meju, and the meju then served as a traditional Korean soybean fermentation starter. The meju were put into large Korean ceramic earthenware jars, called onggi, which are traditional fermentation vessels in Korea (1), with brine and left to further ferment for 8 weeks. After fermentation in the brine, the liquid and meju were separated and the liquid became traditional Korean soy sauce, kanjang. The meju was subsequently crushed and became traditional homemade soybean paste, doenjang. After the separation and subsequent boiling, the traditional soy sauce was aged for 1, 2, 4 or 12 years that are used for this study.
Onggi Tank for fermentation of Korean Traditional Soy Sauce. Raw soy sauce is traditionally boiled for 30 minutes to stop growth of any undesirable microorganisms and to inactivate residual enzymes after filtration or pressing, resulting in stabilization in the flavor and color during storage (2). Recently, ceramic membranes have been used for microfiltration instead of boiling to provide commercial products with constant quality (3). Most cottage manufacturers, especially generational manufacturers in Korea, traditionally S-3
boil raw soy sauce to induce aging for at least 1 year and sometimes for over 10 years in an onggi tank (Figure S-1 in the Supporting Information), ensuring a unique and high-quality product. Onggi is made from ceramic material and is porous, which allows permeation of oxygen, carbon dioxide, water and salt during the fermentation (1). The porosity may provide useful growth conditions for microflora and cause them to be immobilized. The pore size of mesoporous molecular sieves also affects the activity of immobilized enzymes (4, 5). Immobilized cells in the pores of ceramic beads contribute to rapid ethanol fermentation for soy sauce (6). The main advantage of using immobilized cells in soy sauce was well reviewed (7). Recently, it has been reported that the porosity of onggi plays an important role in immobilizing microflora and enzymes and in providing high quality and high reproducibility when used for repeated-batch fermentation of soy sauce (1, 8, 9).
Normalization Effect of Soy Sauce NMR Spectra on PCA Models. In metabolomic studies, normalization is the preprocessing step for 1H NMR spectra that accounts for variations in the overall concentration of samples caused by different dilutions. Therefore, the normalization step is crucial to compensate for differences in the overall concentration. For the 1H NMR spectra of complex biofluids, such as urine and serum, integral normalization is the standard for normalizing NMR spectra (10, 11). The integral normalization procedure divides each signal of a spectrum by the integral of the complete spectrum. Another common procedure is creatinine normalization (12, 13), in which the NMR spectrum is divided by the peak area of the creatinine methyl resonance. In addition, each NMR spectrum can be divided by median spectrum for normalization (14, 15). In the present study, two methods were applied for normalizing the NMR spectra: normalization to total spectral area and normalization by dividing each spectrum by the median spectrum (Figure S-3). S-4
Figure S-1. An Illustration of traditional Korean soy sauce production. (A) Natural inoculation of various microorganisms from the natural environment and rice straw into meju, as a starter for fermentation, for 50 days; (B) Meju with molds and bacteria; (C) Brine fermentation of meju in onggi tank, Korean ceramic earthenware jars for two months; (D) Aging of soy sauce in onggi for at least 1 year, following separation of soybean paste and soy sauce; (E) End product of soy sauce to be commercialized. (F) reveals each process step in producing traditional Korean soy sauce.
S-5
Figure S-2. Two-dimensional (2D) 1H-1H TOCSY NMR spectrum of traditional Korean soy sauce aged for 12 years. The 2D contour plot (B) corresponds to 1D spectrum acquired using NOESYPRESAT pulse sequence (A). Solid rectangles in panel B trace the 1H-covalent network of protons as pairs of symmetric cross-peaks with respect to the diagonal axis. Most metabolites were also assigned by spiking experiments. Key: 1, leucine; 2, isoleucine; 3, valine; 4, lactate; 5, alanine; 6, acetate; 7, arginine; 8, betaine; 9, tyramine; 10, phenylalanine; 11, formate; 14, lysine; 15, γ-aminobutyrate (GABA); 17, ethanol; 18, succinate; 19, tyrosine; 20, proline; 21, choline; 22, uracil; 23, aspartate; 24, methionine; 25, pyroglutamate; 26, glutamate; 27, glycine; 28, trimethylamine; 29, malonate; 30, phosphocholine; 32, oligosaccharide (O1); 33, oligosaccharide (O2); 34, oligosaccharide (O3); 35, glycerol; 36, propionate; 37, hypoxanthine; U1, U2, U3, U4 and U5, unknown.
S-6
Figure S-3. PCA score plots after normalizing 1H NMR spectra of soy sauces to total spectral area (A) and after normalization through division of each NMR spectrum by median spectrum (B). Black boxes and red circles represent commercial soy sauces from CJ and Sempio manufacturers. Green diamonds represent traditional Korean soy sauces aged for 1 year. Filled and open symbols were normalized in the absence and presence of ethanol peaks in the NMR spectra, respectively, demonstrating that ethanol with huge spectral intensity affects the normalization to total spectra area. PCA models were generated after removing ethanol peaks, following normalization.
S-7
Figure S-4. PCA scores plots derived from traditional Korean soy sauces. Unboiled (U) soy sauces were aged for 1 year whereas other soy sauces were aged for 1, 2 and 4 years following boiling for 30 minutes. Scattered plots of unboiled soy sauces reveal more intra-variations than those of boiled soy sauces, resulting in their uncontrolled or inconstant quality.
S-8
Figure S-5. PCA scores (A and C) and loadings (B and D) plots derived from 1H NMR spectra of traditional Korean soy sauces aged for 4 years (4 years), and commercial Korean (CJ) and Japanese (Kikkoman) soy sauces.
S-9
LITERATURE CITED 1.
Lee, K. S.; Lee, Y. B.; Lee, D. S.; Chung, S. K. Quality evaluation of Korean soy sauce fermented in Korean earthenware (Onggi) with different glazes. Int. J. Food Sci. Technol. 2006, 41, 1158-1163.
2.
Luh, B. S. Industrial production of soy sauce. J. Ind. Microbiol. 1995, 14, 467-471.
3.
Li, M. S.; Zhao, Y. J.; Zhou, S. Y.; Xing, W. H.; Wong, F. S. Resistance analysis for ceramic membrane microfiltration of raw soy sauce. J. Membr. Sci. 2007, 299, 122-129.
4.
Yiu, H. H. P.; Wright, P. A.; Botting, N. P. Enzyme immobilisation using siliceous mesoporous molecular sieves. Micropor. Mesopor. Mat. 2001, 44, 763-768.
5.
Yiu, H. H. P.; Wright, P. A.; Botting, N. P. Enzyme immobilisation using SBA-15 mesoporous molecular sieves with functionalised surfaces. J. Mol. Catal. B-Enzym. 2001, 15, 81-92.
6.
Iwasaki, K.; Nakajima, M.; Sasahara, H.; Watanabe, A. Rapid ethanol fermentation for soy-sauce production by immobilized yeast-cells. Agric. Biol. Chem. 1991, 55, 22012207.
7.
Norton, S.; Vuillemard, J. C. Food bioconversions and metabolite production using immobilized cell technology. Crit. Rev. Biotechnol. 1994, 14, 193-224.
8.
Seo, G. H.; Yun, J. H.; Chung, S. K.; P., P. W.; Lee, D. S. Physical properties of Korean earthenware containers affected by soy sauce fermentation use. Food Sci. Biotechnol. 2006, 15, 168-172.
9.
Iwasaki, K. I.; Nakajima, M.; Sasahara, H. Porous alumina beads for immobilization of lactic-acid bacteria and its application for repeated-batch fermentation in soy-sauce production. J. Ferment. Bioeng. 1992, 73, 375-379.
S-10
10.
Bollard, M. E.; Stanley, E. G.; Lindon, J. C.; Nicholson, J. K.; Holmes, E. NMR-based metabonomic approaches for evaluating physiological influences on biofluid composition. NMR Biomed. 2005, 18, 143-162.
11.
Craig, A.; Cloareo, O.; Holmes, E.; Nicholson, J. K.; Lindon, J. C. Scaling and normalization effects in NMR spectroscopic metabonomic data sets. Anal. Chem. 2006, 78, 2262-2267.
12.
Prevot, A.; Martini, S.; Guignard, J. P. Rev. Med. Suisse Romande 2002, 122, 625-630.
13.
van Velzen, E. J. J.; Westerhuis, J. A.; van Duynhoven, J. P. M.; van Dorsten, F. A.; Hoefsloot, H. C. J.; Jacobs, D. M.; Smit, S.; Draijer, R.; Kroner, C. I.; Smilde, A. K. Multilevel data analysis of a crossover designed human nutritional intervention study. J. Proteome Res. 2008, 7, 4483-4491.
14.
Jansen, J. J.; Bro, R.; Hoefsloot, H. C. J.; van den Berg, F. W. J.; Westerhuis, J. A.; Smilde, A. K. PARAFASCA: ASCA combined with PARAFAC for the analysis of metabolic fingerprinting data. J. Chemometr. 2008, 22, 114-121.
15.
Dieterle, F.; Ross, A.; Schlotterbeck, G.; Senn, H. Probabilistic quotient normalization as robust method to account for dilution of complex biological mixtures. Application in H-1 NMR metabonomics. Anal. Chem. 2006, 78, 4281-4290.
S-11
