Protein Threading

Zhanggroup 2003 10 22

Overview

Background protein structure protein folding and designability

Protein threadingCurrent limitations to protein threading

Computational complexity of certain formulations of the protein threading problem

Performance of protein threading systems

References

Protein Structure

Primary, secondary, tertiary structure

Can only refer to the structure of a protein if a particular environment is assumed

solvent environment (aqueous trans-membrane ……)

temperature pH etcDifferent environments yield different

structures or no stable structure at all

Proteins molecules are not completely rigid structures

kinetic energy energetic collisions with solvent molecules

vibrations sidechain conformational changes

flexible sections of the peptide chainThe native tertiary structure of a protein

is thus an average

Protein Folding

Protein folding = searching for a conformation having minimum energy

Factors in protein folding

hydrophobic effects electrostatic charges in residues hydrogen bondingChaperonins,ribosomes

3 stages of folding

denatured unfolded state molten globule state native compact statemost proteins will return to their native

state after forced denaturation

The Protein Folding Problem

Given a proteins amino acid sequence what is its tertiary structure

The protein folding problem is hard

Direct approach :molecular dynamics simulation

Simulate on an atomic level the folding of a single protein molecule

protein = thousands of atomssolvent environment = hundreds to

thousands of molecules => thousands of atoms

Sub-picosecond time scalesrun the simulation for 1-5 secondsWe need another years of Moores law

to make this computation feasible

DesignabilityA protein with a stable native state can

not have another low-energy state nearby in conformational space

A structure is highly designable if its minimum energy state has no low-energy neighbours

Protein Threadinginverse protein folding problem: givena tertiary structure, find an amino acid

sequence that folds to that structureProtein threading: given a library of

possible protein folds and an amino acid sequence find the fold with the

best sequence -> structure alignment (threading)

Evolution depends on designability to preserve function under mutation

Estimate only different protein structures exist in nature (Chothia,1992)

four componentsa library of protein folds (templates)a scoring function to measure the

fitness of a sequence -> structure alignment

a search technique for finding the best alignment between a fixed sequence and structure

a means of choosing the best fold from among the best scoring alignments of a sequence to all possible folds

Scoring Schemes for Sequence->Structure

Alignments

The scoring scheme for a particular threading of a sequence onto a structure measures the degree to which

environmental preferences are satisfied Different amino acid types prefer different

environments e.g. structural preferences: in helix in sheet not exposed to solvent pairwise interactions with neighbouring amino

acids

Formal Statement of the ProteinThreading Problem

C is a protein core having m segments Ci representing a set of contiguous amino acids Let ci be the length of Ci

Sequence a = a1a2…an of amino acids

Current limitations to protein threading

Statistical problems

Definition of neighbor and /or pairwise contact environments:

energetic neighbor ? contact neighbor

Computational Complexity of Finding an Optimal Alignment

The complexity of the protein threading problem depends on whether:

Variable-length gaps are allowed in alignments

the scoring function for an alignment incorporates pairwise interactions between amino acids

Property(I) makes the search space exponential in size to the length of the sequence

Property(Ii) forces a solution to take non-local effects into account

Any protein threading scheme with both properties is NP-complete

(3-SAT Lathrop 1994)

(MAX-CUT Akutsu,Miyano 1999)

Thus all protein threading approaches can be divided

into four groups:

1 no variable length gaps allowed

2 no pairwise interactions considered in scoring function

3 no optimal solution guarantee

4 exponential runtime

Performance of Protein Threading Systems

CASP1(1994) CASP2(1996) CASP3(1998): Critical

Assessment of Structure Prediction meetings

protein threading methods have consistently been

the winners

success depends on structural similarity of target to

known structures

successful even when target sequence and library

sequence have low homology

Much room for improvement in all areas of protein threading e.g.:

algorithms for searching the threading space

reliable biologically accurate scoring functions