3-D Structure Prediction

This is collection of freely accessible web tools, software and databases for the prediction of protein 3-D structure.

Template-base Modeling

Template-based modeling is a class of methods for constructing an atomic-resolution model of a protein from its amino acid sequence. All the tools here accept a protein's amino acid sequence as input, search known 3D structures for appropriate template(s) and generate a 3D model containing the coordinates of the atoms of the protein. When sequence similarity between the target sequence and a protein of known structure is significant (above 30% identity), this process is referred to as (close) homology modeling. In homology modeling, relatively simple sequence comparison methods are applied (e.g. BLAST or PSI-BLAST) in order to find a template, and to generate the alignment between target and template. For such cases you can use automated servers such as ModWeb and SWISS-MODEL. When there is no known structure with significant sequence similarity to the target (below 30% identity), more sophisticated methods are needed to find appropriate templates (if any) and to generate more accurate alignments. This process is referred as distant homology modeling, fold recognition or threading. In general it was shown that in such cases meta-servers that use the results of several servers to produce a consensus prediction, preformed the best

Before you start 3-D structure prediction, check if your protein has more than one domain or if it has disordered regions (see our 2-D structure prediction tool list). Such factors may play significant role in the sensetivity and preformance of many template-based modeling tools.

Meta Servers

open in new window@TOME-2 @TOME-2 allows one to submit an amino acid sequence to embedded software dedicated to find homologous sequences (HHSearch, Fugue, Psi Blast, Sp3) and to build molecular structures (Tito, Scwrl, modeller) according set of score of evaluation tools such as Tito, TCoffee, Verify3D, 3D-Jury
3D-Jury
The meta server provides access and results assessment from various remote predictors including, 3D-PSSM, ESyPred3D, FFAS03, FUGUE, HHpred, Inub, mGenTHREADER, SAM-T02, SAM-T06.
open in new windowGeneSilico

The meta server provides access to various remote and local predictors including : 3D-PSSM, FFAS03, FUGUE, HHpred, Inub, mGenTHREADER, Pcons5, phyre, SAM-T06, SP4 . The Pcons consensus server use to evaluate to which extent the alignments agree with each other and if a particular fold can be singled out.

open in new windowLOMETS

Meta server that include locally-installed threading programs FUGUE, HHpred, SAM-T02, SPARKS, SP3 and more. LOMETS generates the final models using a consensus approach.

open in new windowPcons.net
In Pcons.net a quick database search is done against known protein structures using BLAST and RPS-BLAST. If a significant hit is found, an all-atom model is produced. If no significant hits are found the sequence is submitted to publicly available more advanced predictors such as 3D-Jury, FFAS03, FUGUE, HHpred, Inub, mGenTHREADER, SAM-T02, SP3. The results from these servers are analyzed and assessed for structural correctness and the user is presented with a ranked list of possible models.
 

Sequence-base Tools

These tools utilize exclusively sequence information available about the target and the templates to find similarities.
open in new window3D-JIGSAW
Automated homology modeling server. 3D-JIGSAW can run in automatic mode or interactive mode in which you can select the templates and correct the alignments before submitting the modeling job.
open in new windowCPHmodels Sequence-profile and profile-profile comparison algorithm for sequence alignment and distant homology detection.
open in new windowESyPred3D
Alignments are obtained by using neural networks for combining, weighting and screening the results of several multiple alignment programs. The final three dimensional structure is built using the modeling package MODELLER.
FFAS03 Profile-profile comparison algorithm for sequence alignment and distant homology detection.
HOMCOS A server to predict interacting protein pairs and interacting sites by homology modeling of complex structures
open in new windowM4T Automated homology modeling server that uses a combination of multiple templates and iterative optimization of alternative alignments.
open in new windowModWeb Automated homology modeling server. The final three dimensional structure is built using the modeling package MODELLER (need MODELLER licence key).
PROTEUS2 Web server that integrates several algorithms for signal peptide identification, transmembrane helix prediction, transmembrane β-strand prediction, secondary structure prediction and homology modeling.
open in new window(PS)2
The server uses consensus strategy combining several multiple alignment programs. The final three dimensional structure is built using the modeling package MODELLER (need MODELLER licence key).
open in new windowPudge Protein structure prediction pipeline that designed to integrate different sources of information and to to enable a flexible modeling strategy tailored to the needs of a given problem.
open in new windowSWISS-MODEL Automated homology modeling server. Accessible via the web or from the program DeepView. Interactive optimization of the model is possible with DeepView.
   

Fold recognition

These tools utilize sequence information and predicted structural features in their scoring functions.
open in new window3D-PSSM Sequence-profile comparisons coupled with secondary structure and solvation potential information.
open in new windowAGAPE (PredictProtein) Sequence-profile and profile-profile comparisons using sequence and predicted 1D structure (secondary structure and accessibility).
open in new windowDescFold The server combine 6 descriptors including profile-sequence alignment and secondary structure element alignment
FUGUE Sequence-profile comparisons using profiles derived from the structure-based alignments in the HOMSTRAD.
open in new windowHHpred open in new window(Bioinformatics Toolkit) HMM-HMM comparison coupled with secondary structure scoring.
Inub Hybrid fold recognition, combining sequence derived properties with evolutionary information.
open in new windowmGenTHREADER profile-profile alignments coupled with secondary structure information.
open in new windowModLink+ Sequence-profile comparisons coupled with protein-protein interactions data
open in new windowMUSTER Sequence profile-profile alignment coupled with multiple structural information.
open in new windowphyre Sequence-profile comparisons coupled with secondary structure and solvation potential information.The successor of 3D-PSSM.
open in new windowSAM-T02 Sequence-HMM comparison coupled with predicted secondary structure scoring.
open in new windowSAM-T06 Sequence-HMM comparison coupled with multiple structural predictions.
open in new windowSAM-T08 Sequence-HMM comparison coupled with multiple structural predictions.
open in new windowSPARKS2 Knowledge-based energy score with sequence-profile and secondary structure information.
open in new windowSP3 profile-profile alignments using profiles derived from evolution and from depth-dependent structural alignment of fragments.
open in new windowSP4 Fold recognition by concurrent use of solvent accessibility and residue depth.
open in new windowSP5 Improving protein fold recognition by using predicted torsion angles and profile-based gap penalty.
   

Model Building

open in new windowAL2TS Generate 3-D model for a pre-aligned sequence-template structure.
open in new windowWHAT IF Generate 3-D model for a pre-aligned sequence-template structure. The link here is for the web interface of the open in new windowWHAT IF program. (To run modeling job, look in the server menu for "build/check/repair model").
   

Standalone Applications

open in new windowCABS Versatile reduced representation tool for molecular modeling More Info
open in new windowMODELLER Widely used homology modeling program using the spatial restraints method. Info
open in new windowMolDE
Open-source cross-platform graphical environment for homology modeling.
More Info
   

Databases of Theoretically Determined 3-D Models

The following databases include 3-D models of proteins calculated by comparative modeling. Before you start your own modeling project, try these databases.
open in new windowMODBASE Automatic calculations with ModPipe.
open in new windowPMP Models structures from different resources (CSMP, JCSG, MCSG, NESG, NYSGXRC, JCMM, ModBase, SWISS-MODEL Repository)
open in new windowSWISS MODEL Repository Automatic calculations with SWISS_MODEL.
   

Model Selection and Ranking

Check also our resources for model validation
open in new windowSELECTpro Structure-based model selection method derived from an energy function comprising physical, statistical, and predicted structural terms.
   

Hybrid methods combining template-based and ab initio methods

open in new windowI-TASSER Models are built based on multiple-threading alignments by LOMETS and iterative TASSER simulations.
open in new windowpro-sp3-TASSER Iterative metathreading with the first round selecting models from five individual threading methods and the second round selecting models from short TASSER predicted models.
open in new windowROBETTA Automated structure prediction server. In case that a confident match to a protein of known structure is found, the server use it as a template for homology modeling. If no match is found, structure predictions are made using the de novo Rosetta fragment insertion method.

Loop Prediction

open in new windowArchPRED A template based loop structure prediction web server.
open in new windowFREAD A template based loop structure prediction tool (for Linux platforms)
open in new windowMODLOOP
Web server for automated modeling of loops in protein structures. The server relies on the loop modeling routine in MODELLER that predicts the loop conformations by satisfaction of spatial restraint (need MODELLER licence key).
open in new windowRAPPER

Tool for automated modeling of loops in protein structures. Rapper samples the conformational space using a fine-grained set of φ ψ states while avoiding steric clashes.(Mac OS X, FreeBSD, and Linux)
open in new windowSuperLooper A template based loop structure prediction web server. Loop candidates are selected from a database (LIMP) comprising loops of membrane proteins or, alternatively, from (LIP) containing segments of water-soluble proteins with lengths up to 35 residues.
open in new windowWloop Web server for loop homology modeling.