Table 6

Predictive performance in terms of the AUC on the IEDB El-Manzalawy benchmark.

UPDS

SRDS1

SRDS2


NN-W-P1

0.863

0.699

0.673

NN-W

    0.864

    0.705

    0.676

CTD

0.782

0.639

0.634

LA

0.802

0.645

0.606

5-spectrum

0.748

0.429

0.390


The methods included are NN-W-P1 (the NN-based method including data redundancy step-size rescaling and P1-PSSM encoding), NN-W (the NN-based method including data redundancy step-size rescaling), CTD, LA, and 5-spectrum. The performance values for the latter three methods are taken from the El-Manzalawy publication [25]. The benchmark data sets are UPDS: Unique peptides from the IEDB database, SRDS1: Sequence similarity reduced UPDS data excluding peptides sharing 9 mer subsequences, and SRDS2: Sequence similarity reduced SRDS1 data ensuring maximum more than 80% similarity between pairs of peptides. For each allele, the best performing NN-based method is highlighted in bold and the best performing method is underlined.

Nielsen and Lund BMC Bioinformatics 2009 10:296   doi:10.1186/1471-2105-10-296

Open Data