Search

Additional file 1: S1 Fig. Distribution of the observed and predicted z-coordinates on the cross-validation dataset. S2 Fig. Distribution of the observed and predicted z-coordinates on the independent test dataset. S3 Fig. Precision-recall and ROC curves of the predicted topology on the cross-validation dataset. S4 Fig. Precision-recall and ROC curves of the predicted topology on the independent test dataset. S5 Fig. Distribution of the observed and predicted continuous flexibility on the cross-validation dataset. S6 Fig. Distribution of the observed and predicted continuous flexibility on the independent test dataset. S7 Fig. Precision-recall and ROC curve of the predicted two-state flexibility on the cross-validation dataset. S8 Fig. Precision-recall and ROC curve of the predicted two-state flexibility on the independent test dataset. S9 Fig. Distribution of the observed and predicted φ angles on the cross-validation dataset. S10 Fig. Distribution of the observed and predicted φ angles on the independent test dataset. S11 Fig. Distribution of the observed and predicted ψ angles on the cross-validation dataset. S12 Fig. Distribution of the observed and predicted ψ angles on the independent test dataset. S13 Fig. Precision-recall and ROC curves of the predicted secondary structure on the cross-validation dataset. S14 Fig. Precision-recall and ROC curves of the predicted secondary structure on the independent test dataset. S15 Fig. Distribution of the observed and predicted relative solvent accessibility of monomers on the cross-validation dataset. S16 Fig. Distribution of the observed and predicted relative solvent accessibility of monomers on the independent test dataset. S17 Fig. Distribution of the observed and predicted relative solvent accessibility of protein chains in complexes on the cross-validation dataset. S18 Fig. Distribution of the observed and predicted relative solvent accessibility of protein chains in complexes on the independent test dataset. S19 Fig. Distribution of the observed and predicted difference in relative solvent accessibility between the bound and the unbound form of a protein chain on the cross-validation dataset. S20 Fig. Distribution of the observed and predicted difference in relative solvent accessibility between the bound and the unbound form of a protein chain on the independent test dataset. S1 Table. Performance of several algorithms, including the final method AllesTM, for z-coordinate prediction on the cross-validation and independent test datasets. S2 Table. Protein topology prediction performance. S3 Table. Continuous flexibility prediction performance. S4 Table. Two-state flexibility prediction performance. S5 Table. φ angles performance. S6 Table. ψ angles performance. S7 Table. Secondary structure prediction performance. S8 Table. Secondary structure prediction performance excluding residues situated in transmembrane segments. S9 Table. Monomer solvent accessibility performance. S10 Table. Complex solvent accessibility performance. S11 Table. Change of solvent accessibility performance. S12 Table. Number of sequences in the training, validation, and test datasets across the five folds. Because of the two-step redundancy reduction procedure, the number of proteins in the validation and training parts are dependent on the test data in the particular fold. S13 Table. Parameter values used for the different layer types during random parameter search. S14 Table. Parameter values used during grid search of random forest models. S15 Table. Parameter values used during random search of GBM models.

Background Adoptive immunotherapy offers great potential for treating many types of cancer but its clinical application is hampered by cross-reactive T cell responses in healthy human tissues, representing serious safety risks for patients. We previously developed a computational tool called Expitope for assessing cross-reactivity (CR) of antigens based on tissue-specific gene expression. However, transcript abundance only indirectly indicates protein expression. The recent availability of proteome-wide human protein abundance information now facilitates a more direct approach for CR prediction. Here we present a new version 2.0 of Expitope, which computes all naturally possible epitopes of a peptide sequence and the corresponding CR indices using both protein and transcript abundance levels weighted by a proposed hierarchy of importance of various human tissues. Results We tested the tool in two case studies: The first study quantitatively assessed the potential CR of the epitopes used for cancer immunotherapy. The second study evaluated HLA-A*02:01-restricted epitopes obtained from the Immune Epitope Database for different disease groups and demonstrated for the first time that there is a high variation in the background CR depending on the disease state of the host: compared to a healthy individual the CR index is on average two-fold higher for the autoimmune state, and five-fold higher for the cancer state. Conclusions The ability to predict potential side effects in normal tissues helps in the development and selection of safer antigens, enabling more successful immunotherapy of cancer and other diseases. Electronic supplementary material The online version of this article (doi:10.1186/s12885-017-3854-8) contains supplementary material, which is available to authorized users.