The project is supported in part by Alfred P. Sloan Foundation, NSF Career Award (DBI 1027394), and the National Institute of General Medical Sciences (GM083107, GM084222).

<ol>
	<li>Zhang, Y. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Template-based+modeling+and+free+modeling+by+I-TASSER+in+CASP7.">Template-based modeling and free modeling by I-TASSER in CASP7.</a> Proteins. 69, 108-117 (2007).</li><li>Zhang, Y. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=I-TASSER:+Fully+automated+protein+structure+prediction+in+CASP8.">I-TASSER: Fully automated protein structure prediction in CASP8.</a> Proteins. 77, 100-113 (2009).</li><li>Wu, S., Skolnick, J., Zhang, Y. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Ab+initio+modeling+of+small+proteins+by+iterative+TASSER+simulations.">Ab initio modeling of small proteins by iterative TASSER simulations.</a> BMC Biol. 5, 17-17 (2007).</li><li>Roy, A., Kucukural, A., Zhang, Y. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=I-TASSER:+a+unified+platform+for+automated+protein+structure+and+function+prediction.">I-TASSER: a unified platform for automated protein structure and function prediction.</a> Nat Protoc. 5, 725-738 (2010).</li><li>Wu, S., Zhang, Y. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=LOMETS:+a+local+meta-threading-server+for+protein+structure+prediction.">LOMETS: a local meta-threading-server for protein structure prediction.</a> Nucleic Acids Res. 35, 3375-3382 (2007).</li><li>Zhang, Y., Kihara, D., Skolnick, J. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Local+energy+landscape+flattening:+parallel+hyperbolic+Monte+Carlo+sampling+of+protein+folding.">Local energy landscape flattening: parallel hyperbolic Monte Carlo sampling of protein folding.</a> Proteins. 48, 192-201 (2002).</li><li>Li, Y., Zhang, Y. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=REMO:+A+new+protocol+to+refine+full+atomic+protein+models+from+C-alpha+traces+by+optimizing+hydrogen-bonding+networks.">REMO: A new protocol to refine full atomic protein models from C-alpha traces by optimizing hydrogen-bonding networks.</a> Proteins. 76, 665-676 (2009).</li><li>Zhang, J., Zhang, Y. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=High-resolution+protein+structure+refinement+using+fragment+guided+molecular+dynamics+simulations.">High-resolution protein structure refinement using fragment guided molecular dynamics simulations.</a> , (2011).</li><li>Roy, A., Zhang, Y. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=COFACTOR:+protein-ligand+binding+site+predictions+by+global+structure+similarity+match+and+local+geometry+refinement.">COFACTOR: protein-ligand binding site predictions by global structure similarity match and local geometry refinement.</a> , (2011).</li><li>Zhang, Y., Hubner, I. A., Arakaki, A. K., Shakhnovich, E., Skolnick, J. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=On+the+origin+and+highly+likely+completeness+of+single-domain+protein+structures.">On the origin and highly likely completeness of single-domain protein structures.</a> Proc. Natl. Acad. Sci. U. S. A. 103, 2605-2610 (2006).</li><li>Zhang, Y., Kolinski, A., Skolnick, J. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=TOUCHSTONE+II:+a+new+approach+to+ab+initio+protein+structure+prediction.">TOUCHSTONE II: a new approach to ab initio protein structure prediction.</a> Biophys. J. 85, 1145-1164 (2003).</li><li>Wu, S., Zhang, Y. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=A+comprehensive+assessment+of+sequence-based+and+template-based+methods+for+protein+contact+prediction.">A comprehensive assessment of sequence-based and template-based methods for protein contact prediction.</a> Bioinformatics. 24, 924-931 (2008).</li><li>Zhang, Y., Skolnick, J. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=SPICKER:+a+clustering+approach+to+identify+near-native+protein+folds.">SPICKER: a clustering approach to identify near-native protein folds.</a> J. Comput. Chem. 25, 865-871 (2004).</li><li>Zhang, J., Zhang, Y. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=High-resolution+protein+structure+refinement+using+fragment+guided+molecular+dynamics+simulations.">High-resolution protein structure refinement using fragment guided molecular dynamics simulations.</a> , (2010).</li><li>Zhang, Y. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=I-TASSER+server+for+protein+3D+structure+prediction.">I-TASSER server for protein 3D structure prediction.</a> BMC Bioinformatics. 9, 40-40 (2008).</li><li>Xu, J., Zhang, Y. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=How+significant+is+a+protein+structure+similarity+with+TM-score+=+0.5?.">How significant is a protein structure similarity with TM-score = 0.5?.</a> Bioinformatics. 26, 889-895 (2010).</li><li>Zhang, Y., Skolnick, J. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Scoring+function+for+automated+assessment+of+protein+structure+template+quality.">Scoring function for automated assessment of protein structure template quality.</a> Proteins. 57, 702-710 (2004).</li><li>Barrett, A. J. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Nomenclature+Committee+of+the+International+Union+of+Biochemistry+and+Molecular+Biology+(NC-IUBMB).+Enzyme+Nomenclature.+Recommendations+1992.+Supplement+4:+corrections+and+additions+(1997).">Nomenclature Committee of the International Union of Biochemistry and Molecular Biology (NC-IUBMB). Enzyme Nomenclature. Recommendations 1992. Supplement 4: corrections and additions (1997).</a> Eur J Biochem. 250, 1-6 (1997).</li><li>Ashburner, M. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Gene+ontology:+tool+for+the+unification+of+biology.+The+Gene+Ontology+Consortium.">Gene ontology: tool for the unification of biology. The Gene Ontology Consortium.</a> Nat. Genet. 25, 25-29 (2000).</li><li>Zhang, Y., Skolnick, J. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=TM-align:+a+protein+structure+alignment+algorithm+based+on+the+TM-score.">TM-align: a protein structure alignment algorithm based on the TM-score.</a> Nucleic. Acids. Res. 33, 2302-2309 (2005).</li><li>Xu, D., Zhang, Y. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=RK+Ab+Intio+Protein+Structure+Prediction.">RK Ab Intio Protein Structure Prediction.</a> , (2011).</li><li>Kim, D. E., Chivian, D., Baker, D. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Protein+structure+prediction+and+analysis+using+the+Robetta+server.">Protein structure prediction and analysis using the Robetta server.</a> Nucleic. Acids. Res. 32, W526-W531 (2004).</li><li>Roy, A., Mukherjee, S., Hefty, P. S., Zhang, Y. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=Inferring+protein+function+by+global+and+local+similarity+of+structural+analogs.">Inferring protein function by global and local similarity of structural analogs.</a> , (2011).</li><li>Marchler-Bauer, A., Bryant, S. H. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=CD-Search:+protein+domain+annotations+on+the+fly.">CD-Search: protein domain annotations on the fly.</a> Nucleic. Acids. Res. 32, W327-W331 (2004).</li><li>Finn, R. D. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=The+Pfam+protein+families+database.">The Pfam protein families database.</a> Nucleic. Acids. Res. 38, D211-D222 (2010).</li><li>Zdobnov, E. M., Apweiler, R. <a target="_blank" href="http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?db=PubMed&cmd=Search&doptcmdl=Citation&defaultField=Title+Word&term=InterProScan--an+integration+platform+for+the+signature-recognition+methods+in+InterPro.">InterProScan--an integration platform for the signature-recognition methods in InterPro.</a> Bioinformatics. 17, 847-848 (2001).</li></ol>

No conflicts of interest declared.

The protocol presented above is a general guideline for structure and function modeling using the I-TASSER server. Although, this automated procedure works very well for most of the proteins, human interventions often help to significantly improve the modeling accuracy, especially for the proteins which lack close templates in the PDB library. Users can intervene during the I-TASSER modeling in the following ways: (a) splitting of multi-domain proteins; (b) providing external restraints to improve the structure assembly; and (c) removing templates during the modeling.
Splitting multi-domain protein:
Many long protein sequences frequently contain multiple domains tethered by flexible linker regions, which makes their structure elucidation difficult using both experimental and computational techniques. Nevertheless, as domains are independently folding entities and can perform distinct molecular function; it is desirable to split long multi-domain proteins and model each domain separately. Modeling domains individually will not only speed up the prediction process, but also increases the quality of query-template alignment, resulting in more reliable structure and function predictions.
Domain boundaries in protein sequences can be predicted using freely available external online programs such as NCBI CDD24, PFAM25 or InterProScan26. Also, if LOMETS threading alignments are available for the query protein, domain boundaries can be located by visually identifying long stretches of unaligned residues in the top threading templates (see Step 5.4). These unaligned regions mostly correspond to domain linker regions. If multi-domain templates are already available in the template PDB library with all the query domains aligned, then the query protein can be modeled as full length.
Provide external restraints
The structure assembly simulations in I-TASSER are mainly guided by spatial restraints collected from the LOMETS threading templates. For query proteins that have good threading hit (Norm. Z-score &gt; 1) in the template library, derived spatial restraints are mostly of high accuracy and I-TASSER will generate high resolution structural models for these proteins. Contrarily, for query proteins that have weak or no threading hit (Norm. Z-score &lt; 1), collected spatial restraints often contain errors because of the uncertainty of the template and the alignment. For these protein targets, user-specified spatial information can be very helpful to improve the quality of the predicted model. Users can provide external restraints to the I-TASSER server in two ways:
A. Specify contact/distance restraints
Experimentally characterized inter-residue contacts/distances, for example from NMR or cross-linking experiments, can be specified by uploading a restraint file. An example file is shown in Figure 8, where Column 1 specifies the type of restraint, i.e. "DIST" or "CONTACT". For distance restraint (DIST), columns 2 and 4 contain residue positions (i, j), columns 3 and 5 contain the atom-types in the residue and column 6 specifies the distance between the two specified atoms. For contact restraints (CONTACT), columns 2 and 3 contain the positions (i, j) of residues which should be in contact. The distance between the side chains center of these contacting residue pairs is decided based on observed distances in known structures in PDB. I-TASSER will try to draw these atom pairs close to the specified distance during the structure refinement simulations.
B. Specify a protein structure template
LOMETS threading programs use a representative PDB library to find plausible folds for the query protein. Although using a representative structure library helps to reduce the time required to compute the sequence-structure alignments, it is possible that a good template protein is missed in the library or the template may not have been identified by LOMETS threading programs, even though it is present in the library. In these cases, the user should specify the desired protein structure as the template.
To specify protein structure as an additional template, users can either upload a PDB formatted structure file or specify the PDB ID of a deposited protein structure in PDB library. The I-TASSER will generate the query-template alignment using MUSTER program23 and will collect spatial restraints from both the user specified template and LOMETS templates to guide the structure assembly simulation. Because the accuracy of the LOMETS restraints is different for different targets, the weight of the LOMETS restraints is stronger in easy (homologous) targets than that in hard (non-homologous) targets, which have been systematically tuned in our benchmark training.
Users can also specify their own query-template alignments. The server accepts alignment in two formats: the FASTA format (Figure 9A) and the 3D format (Figure 9B). The FASTA format is standard and described at <a href="http://zhanglab.%20ccmb.med.umich.edu/FASTA/">http://zhanglab. ccmb.med.umich.edu/FASTA/</a>. The 3D format is similar to the standard PDB format (<a href="http://www.wwpdb.org/documentation/format32/sect9.html">http://www.wwpdb.org/documentation/format32/sect9.html</a>), but two additional columns derived from the templates are added to the ATOM records (see Figure 9B):
 
Columns 1-30: Atom (C-alpha only) and residue names for the query sequence. 
Columns 31-54: Coordinates of C-alpha atoms of the query copied from the corresponding atoms in the template. 
Columns 55-59: Corresponding residue number in the template based on alignment 
Columns 60-64: Corresponding residue name in the template
Exclude templates proteins
Proteins are flexible molecules and can adopt multiple conformational states to change their biological activity. For example, structures of many protein kinases and membrane proteins have been solved in both active and inactive conformation. Also presence or absence of bound ligand can cause large structural movements. While all the conformational states of the template are alike for the threading programs, it is desirable to model the query using templates in only one particular state. A new option on the server allows the user to exclude template proteins during structure modeling. This feature would also allow the user to choose the homology level of templates to be used for the modeling. Users can exclude template proteins from the I-TASSER library by:
A. Specifying a sequence identity cutoff
Users can use this option to exclude homologous proteins from the I-TASSER template library. The homology level is set based on the sequence identity cutoff, i.e. the number of identical residue between the query and the template protein divided by the sequence length of the query sequence. For example, if the user types in "70%" in the provided form, all templates proteins which have a sequence identity &gt;70% to the query protein I-will be excluded from the I-TASSER template library.
B. Exclude specific template proteins
Specific template proteins can be excluded from the I-TASSER template library by uploading a list containing PDB IDs of the structures to be excluded. An example file is shown in Figure 10. As the same protein can exist as multiple entries in the PDB library, I-TASSER server will by default exclude the specified templates (in Column1) as well as all other templates from the library that have an identity &gt;90% to the specified templates. Users can also specify a different identity cutoff, e.g. 70%, where all templates with identity &gt;70% to specified template proteins will be excluded.

<ol>
	<li>FASTA formatted amino acid sequence of the protein to be modeled (see, <a href="http://www.ncbi.nlm.nih.gov/BLAST/fasta.shtml">http://www.ncbi.nlm.nih.gov/BLAST/fasta.shtml</a>).</li>
	<li>A personal computer with access to the internet and a web browser.</li>
	<li>Molecular visualizing software, e.g. RASMOL or PYMOL, for analyzing the predicted tertiary structure and functional sites.</li>
</ol>

a protocol for computer-based protein structure and function prediction

Genome sequencing projects have ciphered millions of protein sequence, which require knowledge of their structure and function to improve the understanding of their biological role. Although experimental methods can provide detailed information for a small fraction of these proteins, computational modeling is needed for the majority of protein molecules which are experimentally uncharacterized. The I-TASSER server is an on-line workbench for high-resolution modeling of protein structure and function. Given a protein sequence, a typical output from the I-TASSER server includes secondary structure prediction, predicted solvent accessibility of each residue, homologous template proteins detected by threading and structure alignments, up to five full-length tertiary structural models, and structure-based functional annotations for enzyme classification, Gene Ontology terms and protein-ligand binding sites. All the predictions are tagged with a confidence score which tells how accurate the predictions are without knowing the experimental data. To facilitate the special requests of end users, the server provides channels to accept user-specified inter-residue distance and contact maps to interactively change the I-TASSER modeling; it also allows users to specify any proteins as template, or to exclude any template proteins during the structure assembly simulations. The structural information could be collected by the users based on experimental evidences or biological insights with the purpose of improving the quality of I-TASSER predictions. The server was evaluated as the best programs for protein structure and function predictions in the recent community-wide CASP experiments. There are currently &gt;20,000 registered scientists from over 100 countries who are using the on-line I-TASSER server.

Watch this Scientific Journal Video about A Protocol for Computer-Based Protein Structure and Function Prediction at JoVE.com

A Protocol for Computer-Based Protein Structure and Function Prediction

Guidelines for computer based structural and functional characterization of protein using the I-TASSER pipeline is described. Starting from query protein sequence, 3D models are generated using multiple threading alignments and iterative structural assembly simulations. Functional inferences are thereafter drawn based on matches to proteins with known structure and functions.

Genome sequencing projects have ciphered millions of protein sequence, which require knowledge of their structure and function to improve the ...

a-protocol-for-computer-based-protein-structure-function

Research

JoVE Journal

Biology

68.1K Views.  University of Michigan. Guidelines for computer based structural and functional characterization of protein using the I-TASSER pipeline is described. Starting from query protein sequence, 3D models are generated using multiple threading alignments and iterative structural assembly simulations. Functional inferences are thereafter drawn based on matches to proteins with known structure and functions.