I-TASSER server: new development for protein structure and function predictions

Abstract

The I-TASSER server (http://zhanglab.ccmb.med.umich.edu/I-TASSER) is an online resource for automated protein structure prediction and structure-based function annotation. In I-TASSER, structural templates are first recognized from the PDB using multiple threading alignment approaches. Full-length structure models are then constructed by iterative fragment assembly simulations. The functional insights are finally derived by matching the predicted structure models with known proteins in the function databases. Although the server has been widely used for various biological and biomedical investigations, numerous comments and suggestions have been reported from the user community. In this article, we summarize recent developments on the I-TASSER server, which were designed to address the requirements from the user community and to increase the accuracy of modeling predictions. Focuses have been made on the introduction of new methods for atomic-level structure refinement, local structure quality estimation and biological function annotations. We expect that these new developments will improve the quality of the I-TASSER server and further facilitate its use by the community for high-resolution structure and function prediction.

Figure 1.

The flowchart of the I-TASSER server pipeline consists of three steps: template identification, full-length structure assembly and structure-based function annotation.

Figure 2.

Illustration of submitted sequence and predicted local structure features in the I-TASSER server output. (a) Amino acid sequence in FASTA format; (b) predicted secondary structure; (c) predicted solvent accessibility; (d) predicted normalized B-factor.

Figure 3.

An excerpt of the predicted structure model with global and local accuracy estimations. The structure is visualized in rainbow cartoon by the JSmol applet on the left panel. The estimated local accuracy is shown as a plot on the right panel, which indicates that the N-terminal and the residues between 20 and 30 have relatively higher modeling error while most of other regions are accurate with estimated distance to native smaller than 2 Å in this example.

Figure 4.

The top 10 PDB proteins that are structurally close to the example protein. The query structure and the PDB proteins are shown in cartoon and backbone, respectively. Each of the structural alignments can be visualized interactively by clicking the corresponding radio buttons.

Figure 5.

Illustration of the predicted ligand-binding site, enzyme commission number and active site. The query structure is shown in gray cartoon. (a) The predicted ligand-binding site. The predicted binding ligands and ligand-binding residues are highlighted in yellow-green spheres and blue ball-and-sticks, respectively. For each prediction, two types of complex structures are provided for download, one containing a representative ligand (i.e. the ‘Rep’ link) and the other containing multiple ligands (i.e., the ‘Mult’ link), respectively. (b) The predicted EC number and active site residues are shown in colored ball-and-sticks.