Online Services

●I-TASSER ●QUARK ●LOMETS ●COACH ●COFACTOR ●MUSTER ●SEGMER ●FG-MD ●ModRefiner ●REMO ●SPRING ●COTH ●BSpred ●SVMSEQ ●ANGLOR ●BSP-SLIM ●SAXSTER ●ThreaDom ●EvoDesign ●GPCR-I-TASSER ●BindProf ●ResQ ●IonCom ●STRUM

●TM-score ●TM-align ●MMalign ●NWalign ●EDTSurf ●MVP ●MVP-Fit ●SPICKER ●HAAD ●PSSpred ●3DRobot ●I-TASSER-MR

●BioLiP ●E. coli ●GLASS ●GPCR-HGmod ●GPCR-RD ●GPCR-EXP ●TM-fold ●DECOYS ●POTENTIAL ●RW ●HPSF ●CASP7 ●CASP8 ●CASP9 ●CASP10 ●CASP11

Back to CEthreader homepage

About CEthreader Server

Overview

CEthreader (Contact Eigenvector-based threader) is a template-based protein structure prediction algorithm guided by contact maps. CEthreader converts contact map predicted by ResPRE into a set of single-body Eigenvectors through Eigendecomposition technique, and subsequently performs dynamic programming based on the contact Eigenvector together with secondary structure and sequence profile to identify templates. Combination of contacts, secondary structures, and profiles enables CEthreader to improve the accuracy in template detection compared to traditional profile-based threading.

A flowchart of the CEthreader pipeline is depicted in Figure 1

Methods

Step 1: features preparation. In this step, an in-house MSA (multiple sequence alignment) detection software is utilized for collecting homology sequences from UNICLUST, UNIREF90 and METACLUST databases. The MSA is used for building Henikoff profile, and secondary structure prediction by PSSpred. The residue-residue contacts of query are predicted by ResPRE for residue pairs with sequence separation of ≥5, as well as residue pairs located at the same helix with a sequence separation of 4. The residue pairs are ranked in descending order of the confidence scores of the predicted contacts by ResPRE. The top 2.51L residue pairs predicted to be in contacts are selected to form the final contact map of the query.

Step 2: templates detection. For a given protein with length L, its contact map is an L×L square, binary and symmetric matrix, where residue pairs that are in contacts are designated as 1 and non-contacting residue pairs are set as 0. Eigendecomposition of the contact map is performed, followed by selection of the largest 7 Eigenvalues and corresponding Eigenvectors to calculate the 7-dimensional contact Eigenvector sequences. A scoring function combing contact Eigenvector, secondary structure and profile termss is utilized by semi-global dynamic programming to compare the query sequence with each template in our database. Different templates are then ranked by contact map overlap. Finally, top 5 templates are selected for building full length models.

Step 3: model generation. The final models are built by MODELLER with restraints generated from templates and predicted secondary structure.

CEthreader flowchart
Figure 1. Pipeline of CEthreader.

User Inputs

The user need to paste the FASTA format amino acid sequence to input box, or upload the amino acid sequence of the query protein through browser button.

Figure 2. User inputs.

Content in output page

The outputs of CEthreader include:

(i) predicted secondary structure and contact map;
(ii) top 10 templates and associated alignments;
(iii) top 5 full-length models;

Illustration of outputs

• Predicted secondary structure and contact map:

  
  Figure 3. Predicted secondary structure and contact map.
  

  
  
• CEthreader Top Templates:

  
  Figure 4. CEthreader top templates.
  

  
  
• Full-length Models:

  
  Figure 5. Models.
  

  
  

Standalone package instruction

If you need to build your own CEthreader pipeline, please download the standalone package, and read the readme file.

---

References:

Wei Zheng, Qiqige Wuyun, Yang Li, S. M. Mortuza, Chengxin Zhang, Robin Pearce, Jishou Ruan, Yang Zhang. Detecting distant-homology protein structures by aligning deep neural-network based contact maps. 2019.

Back to the CEthreader homepage