NetMHCpan-4.0: Improved Peptide–MHC Class I Interaction Predictions Integrating Eluted Ligand and Peptide Binding Affinity Data

Vanessa Jurtz, Sinu Paul, Massimo Andreatta, Paolo Marcatili, Bjoern Peters and Morten Nielsen
J Immunol November 1, 2017, 199 (9) 3360-3368; DOI: https://doi.org/10.4049/jimmunol.1700893

Article Figures & Data

Figures

  • FIGURE 1.
    FIGURE 1.

    Visualization of the neural networks with two output neurons used for combined training on BA and EL data.

  • FIGURE 2.
    FIGURE 2.

    Mean performance per MHC molecule measured in terms of AUC for the four methods: BA, EL, BA+EL BA, and BA+EL EL. The methods were evaluated on all BA (BA_all) data and all EL (EL_all) data, including negative peptides derived from source proteins, and on data sets restricted to alleles occurring in BA and EL data sets (BA_shared and EL_shared).

  • FIGURE 3.
    FIGURE 3.

    (A–C) Predicted length preference of selected MHC molecules according to different models. Binding to selected HLA molecules was predicted for 80,000 8–15-mer peptides, and the frequency of peptide lengths in the top 2% of predicted peptides was calculated. (D) Correlation of predicted and observed ligand length for different models. Binding to all HLA alleles present in both the BA and EL data sets was predicted using the four prediction methods for 80,000 8–15-mer peptides. Subsequently, the occurrence of different peptide lengths in the top 2% of predicted peptides for each molecule was calculated, and the correlation coefficient between these frequencies and the length frequencies in the EL data set was calculated.

  • FIGURE 4.
    FIGURE 4.

    EL LOO experiments. (A) Performance per MHC allele of a model trained on all data and a model in which the EL data of a given allele was left out of the training process. (B) Correlation of predicted and observed ligand length for a model trained on all data and the LOO models.

  • FIGURE 5.
    FIGURE 5.

    Sensitivity of different models as a function of the Frank threshold on (A) ELs published by Pearson et al. (17) and (B) T cell epitope data downloaded from IEDB.

  • FIGURE 6.
    FIGURE 6.

    Binding motifs for HLA molecules derived from in vitro BA data using a binding threshold of 500 nM (upper panels), and EL data (lower panels). Binding motif plots were made using Seq2Logo with default parameters (30).

  • FIGURE 7.
    FIGURE 7.

    Motivation for using percentile rank score predictions. Box-plot representation of prediction values for the ligands in the Pearson data set. EL likelihood prediction scores (left panel) and percentile rank values (right panel).

  • FIGURE 8.
    FIGURE 8.

    Sensitivity and specificity performance curves for the NetMHCpan-4.0 EL likelihood predictions. Curves are estimated from a balanced set of ELs from the data set of Pearson et al. (17). The inset shows the complete sensitivity and specificity curves as a function of the percentile rank score. The main plot shows the curves in the high-scoring range for 0–5 percentile scores. Dashed vertical and horizontal lines indicate sensitivity and specificity and the 2% rank score threshold.

  • FIGURE 9.
    FIGURE 9.

    Predictive performance measured in terms of AUC on the Bassani-Sternberg unfiltered EL data sets. Prediction values are assigned to each peptide in a given data set as the lowest percentile rank score/highest prediction score to each of the HLA molecules expressed by the given cell line. The six methods included are EL RNK (NetMHCpan-4.0 EL percentile rank), EL SCO (NetMHCpan-4.0 EL likelihood score), BA RNK (NetMHCpan-4.0 BA percentile rank), BA SCO (NetMHCpan-4.0 BA score), 3.0 RNK (NetMHCpan-3.0 percentile rank), and 3.0 SCO (NetMHCpan-3.0 BA score).

  • FIGURE 10.
    FIGURE 10.

    Predictive performance evaluated in terms of rank of neoantigens identified in four melanoma samples. A rank value of 1 corresponds to the ligand obtaining the highest score (lowest percentile rank) of all peptides from the given sample. Data and performance values for MixMHCpred are from Bassani-Sternberg et al. (14). NetMHCpan-4.0 and NetMHCpan-3.0 are performance values obtained by assigning to each peptide in the given data set the lowest percentile rank score to each of the HLA-A and -B molecules expressed by the given cell line. The values in parentheses for NetMHCpan-4.0 are the predicted percentile rank values. Lowest rank value for each ligand is highlighted in bold.

Back to top

In this issue

Print
Download PDF
Article Alerts
Email Article
Citation Tools

Jump to section