140 Supplementary Tables All supplementary tables are available online via the following link (bioRxiv preprint server): https://www.biorxiv.org/content/10.1101/2021.12.07.471581v1. Supplementary Table S1. Associations between the presences of antibody-bound peptides and patient age by comparison of the first (Q1, <28 years) and fourth age quartiles (Q4, >53 years). Supplementary Table S2. Differentially abundant antibody-bound peptides between male (n=166) and female (n=297) patients with IBD. Supplementary Table S3. Differentially abundant antibody-bound peptides between patients with IBD (full cohort) and healthy individuals (logistic regression analysis, adjusted for the effects of age and gender). Supplementary Table S4. Differentially abundant antibody-bound peptides between patients with CD and healthy individuals (logistic regression analysis, adjusted for the effects of age and gender). Supplementary Table S5. Differentially abundant antibody-bound peptides between patients with UC and healthy individuals (logistic regression analysis, adjusted for the effects of age and gender). Supplementary Table S6. Differentially abundant antibody-bound peptides between patients with CD and an equally sized subset of age- and sex-matched healthy controls (logistic regression analysis, adjusted for the effects of age and gender). Supplementary Table S7. Differentially abundant antibody-bound peptides between patients with UC and an equally sized subset of age- and sex-matched healthy controls (logistic regression analysis, adjusted for the effects of age and gender). Supplementary Table S8. Differentially abundant antibody-bound peptides between patients with CD and UC (logistic regression analysis, adjusted for the effects of age and gender). Supplementary Table S9. Summary of evaluation metrics of three different classification models (CD vs HCs, UC vs HCs and CD vs. UC) using four different (machine learning-based) classification methods. Supplementary Table S10. Beta coefficients and relative importances of antibody-bound peptides for all three classification tasks (CD vs HCs, UC vs. HCs and CD vs. UC). Supplementary Table S11. Summary of evaluation metrics of recursive feature elimination (RFE)- optimized classification models (CD vs. HCs, UC vs. HCs and CD vs. UC) when selecting the top five (5) and top ten (10) antibody-bound peptides. Supplementary Table S12. Beta coefficients and relative importances of the top five (5) and top ten (10) antibody-bound peptides contributing to model classification performance obtained using recursive feature elimination (RFE). Supplementary Table S13. Comparison of areas under the curve (AUCs) of the non-optimized logistic regression with elastic net models, the RFE-optimized top 5 antibody-bound peptidecontaining model and the RFE-optimized top 10 antibody-bound peptide-containing model (using DeLong’s test). Chapter 4
RkJQdWJsaXNoZXIy MjY0ODMw