I-TASSER server for protein 3D structure prediction.

Zhang Y. · 2008

Read I-TASSER server for protein 3D structure prediction. by Zhang Y. (2008) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc2245901.txt.

Introduction

Zhang presented the I-TASSER server in 2008, an automated system predicting three-dimensional protein structure from amino acid sequence. The method had ranked best in the server section of the recent CASP experiment, a blind community test of prediction methods. The online server at the University of Kansas had already answered thousands of modelling requests from over 35 countries. Each query returns up to five full-length models.

A confidence measure called C-score estimates each model's accuracy from template consensus and clustering density. In large benchmark tests, C-score correlated strongly with the true structural similarity score, at 0.91 on that scale. Using a C-score cutoff gave false positive and false negative rates below 0.1 in these tests. The score reliably separates correct from incorrect overall folds. Average prediction error was just 0.08 on that similarity scale. The server is free for academic use.

Catalog record

CREATORZhang Y.
YEAR2008
COLLECTIONPAPERS
FORMATTXT
FILE SIZE29 KB
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC2245901

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

I-TASSER server for protein 3D structure prediction By Yang Zhang Published in: BMC Bioinformatics DOI: 10.1186/1471-2105-9-40 ABSTRACT Abstract Background Prediction of 3-dimensional protein structures from amino acid sequences represents one of the most important problems in computational structural biology. The community-wide Critical Assessment of Structure Prediction (CASP) experiments have been designed to obtain an objective assessment of the state-of-the-art of the field, where I-TASSER was ranked as the best method in the server section of the recent 7th CASP experiment. Our laboratory has since then received numerous requests about the public availability of the I-TASSER algorithm and the usage of the I-TASSER predictions. Results An on-line version of I-TASSER is developed at the KU Center for Bioinformatics which has generated protein structure predictions for thousands of modeling requests from more than 35 countries. A scoring function (C-score) based on the relative clustering structural density and the consensus significance score of multiple threading templates is introduced to estimate the accuracy of the I-TASSER predictions. A large-scale benchmark test demonstrates a strong correlation between the C-score and the TM-score (a structural similarity measurement with values in [0, 1]) of the first models with a correlation coefficient of 0.91. Using a C-score cutoff > -1.5 for the models of correct topology, both false positive and false negative rates are below 0.1. Combining C-score and protein length, the accuracy of the I-TASSER models can be predicted with an average error of 0.08 for TM-score and 2 Å for RMSD. Conclusion The I-TASSER server has been developed to generate automated full-length 3D protein structural predictions where the benchmarked scoring system helps users to obtain quantitative assessments of the I-TASSER models. The output of the I-TASSER server for each query includes up to five full-length models, the confidence score, the estimated TM-score and RMSD, and the standard deviation of the estimations. The I-TASSER server is freely available to the academic community at http://zhang.bioinformatics.ku.edu/I-TASSER. BACKGROUND Protein structure prediction refers to the effort of generating 3-dimensional models from amino acid sequences using computer algorithms. However, structure modeling processes often involve human interventions because the human-expert knowledge combined with biochemical information (function, mutagenesis, catalytic residues, etc.) could help in both structural assembly and model selection [1,2]. Nevertheless, the development of fully-automated algorithms has the advantage in the potential application on proteome-scale structure predictions [3,4]. Especially, it allows non-experts to generate structural models for their own sequences through Internet services. In the recent community-wide blind experiment, CASP7, I-TASSER (as 'Zhang-Server') generated the best 3D structure predictions among all automated servers. The average GDT_TS [5] or TM-score [6] of all 124 targets/domains is at least 5% higher than the second best server and comparable with the best human-expert predictions [7]. Since the first public release in November 2006, the I-TASSER server has generated structure predictions for thousands of modeling requests from various laboratories in the world. We have been frequently asked by the users about how the quality of the I-TASSER models should be annotated because this will essentially decide how they will exploit the predictions in their research. The general idea of the modeling quality estimation of 3D models has been pursued by a number of authors [8-10], which merges as a new research topic of "model quality assessment programs" (MQAP) [11] and is assessed in the recent CASP7 experiment in the category of QA [12]. In this work, we introduce the on-line setting of the I-TASSER server and develop a confidence scoring system which can provide the users with a simple and reliable assessment of the I-TASSER models. Different from most of the MQAP programs that assess models purely based on the structure of the final models, the confidence scoring function developed here incorporates the information and parameters of the modeling simulations. IMPLEMENTATION I-TASSER method I-TASSER is a hierarchical protein structure modeling approach based on the secondary-structure enhanced Profile-Profile threading Alignment (PPA) [13] and the iterative implementation of the Threading ASSEmbly Refinement (TASSER) program [14]. The detail of the I-TASSER method has been described in [15,16]. Here we give a brief overview of the method. The target sequences are first threaded through a representative PDB structure library (with a pair-wise sequence identity cut-off of 70%) to search for the possible folds by four simple variants of PPA methods, with different combinations of the hidden Markov model [17] and PSI-BLAST [18] profiles and the Needleman-Wunsch [19] and Smith-Waterman [20] alignment algorithms. The continuous fragments are then excised from the threading aligned regions which are used to reassemble full-length models while the threading unaligned regions (mainly loops) are built by ab initio modeling [21]. The conformational space is searched by replica-exchange Monte Carlo simulations [22]. The structure trajectories are clustered by SPICKER [23,24] and the cluster centroids are obtained by the averaging the coordinates of all clustered structures. To rule out the steric clashes on the centroid structures and to refine the models further, we implement the fragment assembly simulation again, which starts from the cluster centroid of the first round simulation. Spatial restraints are extracted from the centroids and the PDB structures searched by the structure alignment program TM-align [25], which are used to guide the second round simulation. Finally, the structure decoys are clustered and the lowest energy structure in each cluster is selected, which has the Cα atoms and the side-chain centers of mass specified. Pulchra [26] is used to add backbone atoms (N, C, O) and Scwrl_3.0 [27] to build side-chain rotamers. If any region with >80 residues has no aligned residues in at least two strong PPA alignments of Z-score > Z0 (see below), the target will be judged as a multiple domain protein and domain boundaries are automatically assigned based on the borders of the large gaps. I-TASSER simulations will be run for the full chain as well as the separate domains. The final full-length models are generated by docking the model of domains together. The domain docking is performed by a quick Metropolis Monte Carlo simulation where the energy is defined as the RMSD of domain models to the full-chain model plus the reciprocal of the number of steric clashes between domains. The goal of the docking is to find the domain orientation that is closest to the I-TASSER full-chain model but has the minimum steric clashes. This procedure does not influence the multiple domain proteins which have all domains completely aligned by the PPAs. C-score The C-score of the I-TASSER models is defined as [formula: (1)C-score=ln(MMtot⋅1〈RMSD〉⋅∏i=14Z(i)∏i=14Z0(i))] where M is the multiplicity of structures in the SPICKER cluster; Mtot is the total number of the I-TASSER structure decoys used in the clustering; ⟨RMSD⟩ is the average RMSD of the decoys to the cluster centroid; Z(i) is the highest Z-score (the energy to mean in the unit of standard deviation) of the templates by the ith PPA threading program and Z0(i) is a program-specified Z-score cutoff for distinguishing between good and bad templates, i.e. Z0(1) = 7.0, Z0(2) = 8.5, Z0(3) = 8.0, Z0(4) = 10.5. The first two factors of Equation 1 account for the degree of structure convergence in the SPICKER clustering, which correlates with the consistency of the external restraints and the inherent I-TASSER potential. The third factor accounts for the quality

Read the complete text