Welcome to 3D-JIGSAW (version 3.0)

This server builds three-dimensional models for proteins based on homologues of known structure.

Templates are identified using HMM (Soding, 2005). The returned alignments are used to built the models. All models are preselected using POPULUS ENERGY. Gaps and missing residues are closed and filled using POPULUS REPAIR. Finally all models are recombined using the basic POPULUS approach.

In the POPULUS approach, we use a genetic algorithm (GA) as our conformational space search engine. GAs are well-known to be powerful search and optimisation techniques, and have been used previously in a number of protein modelling efforts ranging from ab initio folding to model-building by homology. In contrast to refinement protocols based on molecular dynamics, Monte Carlo and knowledge-based techniques, which usually start with models created by a single modelling technique, our method uses multiple initial models from a variety of modelling approaches.

After 10 rounds or conversion, all models are ranked using a fine and coarse energy function weighted according to the highest SEQID found. The top five models are returned.

There are three modes avaiable. In the automatic mode the models used for populus are selected automatically. In the interactive mode you can select the models recombined, according to their coverage, sequence id and domains. In the upload mode, the uploaded models are recombined.

If you are not happy with the results from an automatic submission, feel free to resubmit your job using the interactive mode. Here you can select the templates to use, the best coverage, look at functions and realign the query and template sequence.

Submit a modelling job here.

Get a basic version of POPULUS software and request a password here.