This list of web severs is obtained from the published paper “Evaluation of disorder predictions in CASP9” by Andriy Kryshtafovych.
List of web servers and the methods descriptions (copy from table III of Kryshtafovych’s paper). Please read the paper for details of methods and performance.
- PrDOS2 (http://prdos.hgc.jp/cgi-bin/top.cgi)
SVM algorithm based on sequence profiles combined with a template-based predictor.
- DisoPred3C (http://bioinf.cs.ucl.ac.uk/disopred)
SVM trained on high-resolution X-ray structures. Uses profiles from 15 positions around each residue as an input vector.
- MULTICOM-refine (http://casp.rnet.missouri.edu/predisorder.html)
One-Dimensional Recursive Neural Network (1D-RNN). Predicts the disorder probability of each residue along a protein sequence taking as input the sequence profile, predicted secondary structure, and solvent accessibility.
- Zhou-Spine-D (http://http://sparks.informatics.iupui.edu/SPINE-D/)
Two-layered Neural Network followed by a filter. The input features include residue-level and window-level information calculated from amino acid sequence, seven representative physical parameters, PSI-BLAST profile, predicted secondary structure and solvent accessibility torsion-angle fluctuation.
- Zhou-Spine-DM (http://http://sparks.informatics.iupui.edu/SPINE-DM/)
Meta approach that employs a two-layer Neural Network with a filter. Combines input from six disorder predictors: VSL2, DISOPred2, DisPro1.0, IUPred and SPINE-D (above).
- CBRC_Poodle (http://mbs.cbrc.jp/poodle/poodle-i.html)
SVM integrating three own SVM predictors: Poodle-S and Poodle- L specialized in short and long disorder regions, respectively, and Poodle-W targeting unfolded protein prediction.
- biomine_dr_pdb, biomine_dr_pdb_c (http://biomine-ws.ece.ualberta.ca/MFDp.html)
Two meta + SVM approaches. Combine predictions from DISOPred2, DISOclust and IUPred, and additionally use predicted secondary structure, solvent accessibility, B-factors and backbone dihedral torsion angles in SVM learning.
- GSmetaDisorderMD, GSmetaserver (http://iimcb.genesilico.pl/metadisorder/)
Two Genetic Algorithms combining predictions from GSmetaDisorder (consensus of 13 DR servers) and GSmetaDisorder3D (consensus of missing residues in multiple sequence alignments produced by fold recognition methods). The two methods use different weight optimization scores.
- OnD-CRF (http://babel.ucmp.umu.se/ond-crf/)
Machine learning technique based on Conditional Random Fields. Uses only sequence and predicted secondary structure as inputs.
- Mason (http://www.cs.gmu.edu/~mlbio/svmprat)
SVM method integrating flexible window-based profile kernels and predicted secondary structure.
- McGuffin (as DISOClust on all CASP9 3D server models) (http://www.reading.ac.uk/bioinf/DISOclust/)
DISOClust server was tested in CASP9 as a server group (IntFOLD- DR) and a human-expert group (McGuffin). Both groups exploited the same approach based on conservation of ordered residues within multiple structures. The McGuffin group used consensus of all 3D server models submitted to CASP, while IntFOLD-DR – only of those available from the in-house nFOLD4 method.