Supporting Information for:
Glycomic Signatures on Serum IgGs forPrediction of Postvaccination Response
Jing-Rong Wang1,†, Wen-Da Guan2,†, Lee-Fong Yau1, Wei-Na Gao1, Yang-Qing Zhan2, Liang Liu1, Zi-Feng Yang2,*, Zhi-Hong Jiang1,2,*
1State Key Laboratory of Quality Research in Chinese Medicine, Macau Institute for Applied Research in Medicine and Health, Macau University of Science and Technology, Macau, China.
2State Key Laboratory of Respiratory Disease, National Clinical Research Center for Respiratory Disease, First Affiliated Hospital of Guangzhou Medical University, Guangzhou, Guangdong, China.
†These authors contributed equally to the work.
*To whom correspondence should be addressed. E-mail: (Z.-H. Jiang); (Z.-F. Yang);
- Analysis on the correlation between the baseline HI titer and the antibody response.
We analyzed the relationship between the baseline HI titer and the antibody response for H1 and B vaccine strains. For FluB, the point-biserial correlation coefficient of the baseline HI titers and antibody response (HI titer at day 14/HI titerat day 0) is 0.5639, and the p-value (0.0027) is less than 0.01, suggesting that these two attributes are correlated. When we use the fold changes to predict the sample’s response in the support vector machine (SVM) method with 5-fold cross-validation, the accuracy and the true positive rate of classification and prediction are only 73.1% and 61.5% respectively. Hence, baseline attribute cannot be adopted for the prediction of influenza vaccination.
For H1N1, the point-biserial correlation coefficient of the initial titers (day 0) and the antibody fold changes (Day 14/Day 0) is 0.5336, and the p-value (0.005) is less than 0.01, showing that these two attributes are correlated. However, when we use baseline antibody to predict the sample’s classes in the support vector machine (SVM) method with 5-fold cross-validation, the accuracy and the true positive rate of classification and prediction are only 67.5% and 68.4% respectively. Hence, the baseline HI titers cannot be adopted for the prediction of influenza vaccination.
Figure S1. Relationship between initial titers (day 0) and fold changes (Day 14/day 0) in HI titer for B strain (A) and H1 strain (B) after vaccination.
- Analysis on the correlation between the baseline HI titer and the glycosylation profile
To examine whether the different Fc-glycosylation profiles between responders and nonresponders can be attributed to difference in baseline titers, we analyzed the correlation between the relative levels of predictive glycomic signatures for antibody responses. Consequently, weak correlations for 2 out of 12 glycoform markers of H1 strain (IgG1-G1FS and IgG1-G1N-a) were observed (Supporting information, Figure S3), whereas no potential correlations for glycomic signatures for FluB was detected (Figure S2), indicating that the different baseline glycosylation profiles between responders and nonresponders are not simply correlated with baseline antibody titer, which was further supported by PLS-DA and OPLS-DA analyses of global glycosylation profiles based on the initial antibody titers.
Figure S2. Correlation between baseline HI titer for Flu B and relative level of predicative glycoform markers for the vaccination response to Flu B.
Figure S3.Correlation between baseline HI titer for H1N1 and relative level of predictive glycoform markers for the vaccination response to H1N1.