Fold Prediction Using SVM, Chris Ding

Protein Fold Prediction Data sets used in the paper

"Multi-class Protein Fold Recognition Using Support Vector Machines and Neural Networks", by Chris Ding and Inna Dubchak

Feature vectors (parameter vectors) computed from sequences:

Data format

Amino Acids Composition: training set , testing set

Predicted secondary structure: training set , testing set

Polarity: training set , testing set

Polarizability: training set , testing set

Hydrophobicity: training set , testing set

Van der Waals volume: training set , testing set

Fold names, protein names, and their association

Fold name and index: file
Training protein name to fold index identification file
Testing protein name to fold index identification file

Sequences for the proteins
(Provided by Hong-bin Shen of Shanghai Jiao Tong Univeristy)

Sequences for the training proteins file
Sequences for the test proteins file

(Only sequences of the proteins of the 27 folds used in the paper are provided. Thera are
311 proteins for the training set; 383 proteins for the testing set)