Fold Recognition

Michael Tress
ProteinDesign Group.
Centro Nacional de Biotecnologia (C.N.B.- C.S.I.C.)

Tlf: +34-91-5854570. Fax: +34-91-5854506.
Madrid-Jan-05

YOU'LL FIND HERE

1.- A short Introduction
2.- Prior STEPS .
3.- Some PROGRAMS available for fold recogntion.
4.- An EXAMPLE of a fold recognition procedure.

Introduction

 :::: [BACK TO TOP]


                Fold recogntion and "threading" methods are 3D structure prediction techniques that can be used when the twilight zone is reached (meaning there is less than 25% of identity between 2 proteins in a pair-wise alignment). In this case the homology modelling approach is often unreliable. Threading methods take the target sequence and evaluate its structural fit against different known folds, other fold recogntion methods use a variety of techniques to go much deeper into the twilight zone than sequence search methods. The strategies use vary between fold recognition programs. They might use for instance secondary structure coincidence or accesibility, or solvatation energy, etc.

            In general, methods of protein fold recognition attempt to detect similarities between 3D features in proteins that show no sequence similarity. Unlike sequence-only methods, most fold detection methods use additional information from 3D data. So these methods are often predicting how well a the target sequence fits each fold.




Steps to follow prior to fold recognition

.:::: [BACK TO TOP]


Features to consider Method
1.- First:

  • We need to check that there isn't a homologous protein with known structure.
  • If there is a homologous structure it is better to use homology modelling.
    		•  To answer this question:

  • Run a BLAST search against the PDB database using the target sequence.
    		•
    2.- Does the secondary structure and accessibility of the protein help to evaluate the fold recognition results? Check 1D characteristics


    Some Fold Recognition programs:

    		:::: [BACK TO TOP]


    Consensus Methods
    Servers
    1. 3D-pssm (ICNET). Based on sequence profiles, solvatation potentials and secondary structure (not recently updated).
    2. TOPITS (PredictProtein server) (EMBL). Based on coincidence of secondary structure and accesibility. This server also returns predictions for both of these.
    3. UCLA-DOE Structure Prediction Server (UCLA). Executes various fold recognition programs and reports a consensus.
    4. 123D+. Combines substitution matrix, secondary structure prediction, and contact capacity potentials.
    5. FFAS03 (Burnham Institute). Based on profile-profile matching algorithms of the query sequence with sequences from clustered PDB database.
    6. mGenThreader (UCL) Secondary structure prediction from PSIPRED at the same site.
    7. Fugue 2.0
    8. Raptor - you have to send an email to warn them first
    9. Sam-T02 Based on Markov models of alignments of crystalized proteins.
    10. ROBETTA If ROBETTA cannot create a model by homology modelling it tries to use ab initio approaches to create a fold.


    Other Useful Servers

    CA-GEN
    A server that can generate CA trace for models from any fold recognition (or other) alignment. Combine the output with MaxSprout to generate all atom models.

    Fold to 3D Model Converter
    A server that will convert alignment style models to actual PDB type 3D models

    MaxSprout
    A server that will add side chains to CA traces

    SQUARE
    A server that will check how good your homology modelling or fold recogniton alignment is (CNB, Madrid).



    A Procedure Example


    		:::: [BACK TO TOP]
    Fold Recogntion Procedure EXAMPLE: HYPOTHETICAL FLAVIN REDUCTASE