Epitope Predictions Indicate the Presence of Two Distinct Types of Epitope-Antibody-Reactivities Determined by Epitope Profiling of Intravenous Immunoglobulins
Figure 1
Data set preparation and computational workflow for the prediction of epitope-antibody-reactivities (EAR) determined for IVIG antibodies.
Rectangles represent groups of peptides (numbers in each group are indicated), boxes with rounded corners indicate the applied classification approaches. 1All peptides printed on the microarrays 2Removal of false positive (binding) peptides (e.g. those reactive with secondary antibodies) 3Separation of peptide set according to signal intensities of EAR into non-binders, binders and unassigned peptides 4Classification approach ML-advanced = machine learning with an ensemble classifier 5Number of peptides predicted to be non-binding/binding, separated into those predicted correctly (underlined) and incorrectly 6Classification approach PWM = position weight matrix 7Classification approach ML-simple = simplified machine learning using human-understandable attributes 8Capital letters A–H indicate subsets of peptides assigned in supplementary information table S1 and explained there in the legend.