Figure 2
From: Prediction of mono- and di-nucleotide-specific DNA-binding sites in proteins using neural networks
Prediction performance of mononucleotide-specific contacts using various descriptors. (a) Overall performance (b) performance classified by secondary structure (c) performance classified by solvent accessibility. All performance estimates are in the percentage units of area under the ROC curve (AUC). Abbreviation: SS: Single sequence information. PSSM: prediction using PSSM, SASG: prediction using single sequence, solvent accessibility and secondary structure of target residue, and global amino acid composition of the protein chain. PASG: prediction using PSSM, solvent accessibility and secondary structure of target residue, and global amino acid composition of the protein chain.