Proteins

Proteins control the biological functions of cellular organisms e.g. metabolism, blood clotting, immune system

Building blocks – amino acidsamino acids

amino group (NH2), carboxyl group (COOH), side chain R

The Protein Data Bank

Protein sequence and structure

Protein alphabet consists of 20 amino acids

Sequence view Structure view

ADKELKFLVVDDFSTMRRIV.....

Protein structure and function

Function is determined by 3D shape/structure

ThrombinFacilitates blood clotting

HirudinAnticoagulant

(blocks active site)

Protein structure and function

Structure conserves better evolution information

1MBC: VLSEGEWQLVLHVWAKVE.....

2FAL: XSLSAAEADLAGKSWAPV.....

Myoglobin family

Structural Bioinformatics

Pairwise alignment algorithms DALI (Holm and Sander, Journal of Molecular Biology, 1993) LOCK (Singh and Brutlag, ISMB, 1997) CE (Shindyalov and Bourne, Protein Engineering, 1998) SSM (Krissinel and Henrick, Acta Cryst., 2004) Ye et al. JBCB, 2004

Multiple alignment algorithms Gerstein and Levitt, ISMB, 1996: Iterative dynamic programming SSAP (Orengo and Taylor, Methods Enymol., 1996): Two-level DP Leibowitz et al., ISMB, 1999): Geometric hashing CE-MC (Guda et al., PSB, 2001) MAMMOTH (Lupyan et al., Bioinformatics, 2005) MAPSCI (Ye at al., WABI, 2006)

Structural Bioinformatics

Homology detection Hidden Markov models (Jaakola et al., JCB, 2000) Spectrum, Mismatch kernel (Leslie et al., Bioinformatics, 2002) Structure kernel (Qiu et al., Bioinformatics, 2007)

Protein structure prediction Jones and Hadley, Bioinformatics: Sequence, structure and databanks. 2000. FUGUE (Shi et al., J. Mol. Biol., 2001) SCOP (Andreeva, Nucleic Acids Res., 2004)

Protein docking Shoichet et al., J. Comput. Chem., 1992. Choi et al., WABI, 2004. Wang et al., PSB, 2005. Sousa et al., Proteins, 2006.

Pairwise Structure Alignment Given two proteins represented by the Cα atoms (backbone)

f ind 3D transformation that superimposes a large number of the C α atoms

ensure that overall distance between matched pairs is as small as possible

Trade-off between number of matches and total distance between

Pairwise Structure Alignment Ye et al. JBCB 2004

Uses orientation independent representation of proteins based on the fact that Cα atoms are ~4 Ǻ apart

Pairwise Structure Alignment Ye et al. JBCB 2004

The protein is represented as a sequence of angle triplets{(α1, β1, γ1), (α2, β2, γ2), …, (αn, βn, γn) }

Pairwise Structure Alignment Ye et al. JBCB 2004

Compute a local alignment based on angle representation

Find maximal subset of runs with similar transformation matrices

Pairwise Structure Alignment Ye et al. JBCB 2004

The main algorithm Compute the angle based representation Align the angle based representation Identify runs with similar transformation matrices Compute initial structural alignment Refine the alignment iteratively

Running time is ~(m+n)2 where m, n are the protein lengths

Multiple Structure Alignment Given a set of proteins represented by the Cα atoms (backbone)

find a simultaneous alignment of all structures

find a consensus structure that represents all of them

Multiple Structure Alignment The main algorithm

find initial consensus structure (one of the given proteins) pairwise align the consensus and each of the proteins merge the pairwise alignments from previous step recompute the consensus protein; repeat from step 2

Merging the pairwise alignments similar to sequence case

P1 = BBCA, P2 = CBBA, P3 = BCCA

P1: -BBCA, P1:= BBCA P: -BBCA

P2: CBB-A, P3:= BCCA P: CBB-A

P: -BCCA

Multiple Structure Alignment

Computation of consensus structure (after merging alignments)

Multiple Structure Alignment

Algorithm flowchart