Towards the prediction of non-peptidic epitopes
Table 4
Most important fingerprint features for the prediction of T cell epitopes of the fatty acid derivatives (cluster 2).
The fingerprint feature IDs correspond to Fig 8. The corr. p-value is based on the hypothesis (H0), that the feature count is equally distributed in the epitopes and the background. For explanation of other feature-specific metrics see Methods. For those features where no examples are present in the background dataset, the fold-enrichment and mean count difference cannot be computed.