Figure 4. Analysis of amino acid enrichment in IVIG “binding” versus “non-binding” peptides for all, the 1^st degree classifiable and 1^st degree unclassifiable peptides of the training set.

Amino acid propensities for the indicated groups were determined by dividing the frequency of amino acids in “binding” peptides (recognized by IVIG) by the frequency of amino acids in non-binding peptides. Results are sorted by decreasing epitope propensity assigned to the whole training set group. A propensity score >1 means that an amino acid is more likely to occur in the “binding” peptides, a score <1 more frequent in the “non-binding” ones, respectively. The analysis is position-independent.