Chapter 9 Superposition and Dynamic Programming 1 Chapter 9 Superposition and dynamic programming Most methods for comparing structures use some sorts.

Slides:



Advertisements
Similar presentations
Chapter 4 Euclidean Vector Spaces
Advertisements

More Vectors.
3D Geometry for Computer Graphics
Medical Image Registration Kumar Rajamani. Registration Spatial transform that maps points from one image to corresponding points in another image.
Transformations We want to be able to make changes to the image larger/smaller rotate move This can be efficiently achieved through mathematical operations.
PCA + SVD.
Mapping: Scaling Rotation Translation Warp
Image Indexing and Retrieval using Moment Invariants Imran Ahmad School of Computer Science University of Windsor – Canada.
HCI 530 : Seminar (HCI) Damian Schofield. HCI 530: Seminar (HCI) Transforms –Two Dimensional –Three Dimensional The Graphics Pipeline.
Linear Transformations
Protein Structure Alignment Human Myoglobin pdb:2mm1 Human Hemoglobin alpha-chain pdb:1jebA Sequence id: 27% Structural id: 90% Another example: G-Proteins:
Chapter 4.1 Mathematical Concepts
Chapter 5 Orthogonality
Iterative closest point algorithms
Chapter 4.1 Mathematical Concepts. 2 Applied Trigonometry Trigonometric functions Defined using right triangle  x y h.
CSCE 590E Spring 2007 Basic Math By Jijun Tang. Applied Trigonometry Trigonometric functions  Defined using right triangle  x y h.
Ch. 4: Velocity Kinematics
3-D Geometry.
Chapter 3 Determinants and Matrices
Introduction to Robotics Lecture II Alfred Bruckstein Yaniv Altshuler.
3D Geometry for Computer Graphics
Rotations and Translations. Representing a Point 3D A tri-dimensional point A is a reference coordinate system here.
6 1 Linear Transformations. 6 2 Hopfield Network Questions.
Screw Rotation and Other Rotational Forms
Model Database. Scene Recognition Lamdan, Schwartz, Wolfson, “Geometric Hashing”,1988.
Reflection symmetry If you can draw a line through a shape so that one half is the mirror image of the other then the shape has reflection or line symmetry.
CS 450: Computer Graphics 2D TRANSFORMATIONS
VECTORS AND THE GEOMETRY OF SPACE 12. VECTORS AND THE GEOMETRY OF SPACE A line in the xy-plane is determined when a point on the line and the direction.
Separate multivariate observations
Velocities and Static Force
Point set alignment Closed-form solution of absolute orientation using unit quaternions Berthold K. P. Horn Department of Electrical Engineering, University.
Rotations and Translations
CSE554AlignmentSlide 1 CSE 554 Lecture 8: Alignment Fall 2014.
CSE554Laplacian DeformationSlide 1 CSE 554 Lecture 8: Laplacian Deformation Fall 2012.
Geometric Transformation. So far…. We have been discussing the basic elements of geometric programming. We have discussed points, vectors and their operations.
1 Preview At least two views are required to access the depth of a scene point and in turn to reconstruct scene structure Multiple views can be obtained.
Elementary Linear Algebra Anton & Rorres, 9th Edition
CSE554AlignmentSlide 1 CSE 554 Lecture 5: Alignment Fall 2011.
Course 12 Calibration. 1.Introduction In theoretic discussions, we have assumed: Camera is located at the origin of coordinate system of scene.
Rotations and Translations
1 MAC 2103 Module 6 Euclidean Vector Spaces I. 2 Rev.F09 Learning Objectives Upon completing this module, you should be able to: 1. Use vector notation.
6 1 Linear Transformations. 6 2 Hopfield Network Questions The network output is repeatedly multiplied by the weight matrix W. What is the effect of this.
A 3D Model Alignment and Retrieval System Ding-Yun Chen and Ming Ouhyoung.
Spatial Descriptions and Transformations Sebastian van Delden USC Upstate
INF380 - Proteomics-101 INF380 – Proteomics Chapter 10 – Spectral Comparison Spectral comparison means that an experimental spectrum is compared to theoretical.
Conformational Space.  Conformation of a molecule: specification of the relative positions of all atoms in 3D-space,  Typical parameterizations:  List.
Elementary Linear Algebra Anton & Rorres, 9th Edition
Protein Strucure Comparison Chapter 6,7 Orengo. Helices α-helix4-turn helix, min. 4 residues helix3-turn helix, min. 3 residues π-helix5-turn helix,
CSE554AlignmentSlide 1 CSE 554 Lecture 8: Alignment Fall 2013.
CHAPTER 5 SIGNAL SPACE ANALYSIS
Course14 Dynamic Vision. Biological vision can cope with changing world Moving and changing objects Change illumination Change View-point.
MINRMS: an efficient algorithm for determining protein structure similarity using root-mean-squared-distance Andrew I. Jewett, Conrad C. Huang and Thomas.
Affine Geometry.
CSCE 552 Fall 2012 Math By Jijun Tang. Applied Trigonometry Trigonometric functions  Defined using right triangle  x y h.
Basic Theory (for curve 01). 1.1 Points and Vectors  Real life methods for constructing curves and surfaces often start with points and vectors, which.
Lecture 11 CS5661 Structural Bioinformatics – Structure Comparison Motivation Concepts Structure Comparison.
Topics in bioinformatics CS697 Spring 2011 Class 12 – Mar Molecular distance measurements Molecular transformations.
Lab Meeting 10/08/20041 SuperPose: A Web Server for Automated Protein Structure Superposition Gary Van Domselaar October.
INTRODUCTION TO DYNAMICS ANALYSIS OF ROBOTS (Part 1)
Find the optimal alignment ? +. Optimal Alignment Find the highest number of atoms aligned with the lowest RMSD (Root Mean Squared Deviation) Find a balance.
Sect. 4.5: Cayley-Klein Parameters 3 independent quantities are needed to specify a rigid body orientation. Most often, we choose them to be the Euler.
CSE 554 Lecture 8: Alignment
Lecture 16: Image alignment
Fitting Curve Models to Edges
FP1 Matrices Transformations
Solution of Equations by Iteration
Numerical Analysis Lecture 16.
Superposition and Transposition
The Pinhole Camera Model
Presentation transcript:

Chapter 9 Superposition and Dynamic Programming 1 Chapter 9 Superposition and dynamic programming Most methods for comparing structures use some sorts of superposition and dynamic programming We have two structures A and B with elements in a given order A = A 1 A 2... A m and B = B 1 B 2... B n An equivalence for (A, B) is a set of pairs of elements (A i1, B j1 ) (A i2, B j2 )... (A ir, B jr ) If i 1 <i 2 <...<i r and j 1 <j 2 <...<j r then the set of pairs is an alignment We shall look at methods for finding ”good” - equivalences - alignments

Chapter 9 Superposition and Dynamic Programming 2 Superposition Determine how ”good” an equivalence is: Put the structures on top of each other, and see how good the equivalent pairs fit each other in the space, when translation and rotation are allowed If the geometry of the structures are not changed, this is called rigid body superposition Let B be fixed in space Transform A with a transformation T over B We can measure how good T is by giving a score to the resulting superposition by measuring the root mean square deviation (RMSD) between the equivalent pairs of (transformed ) A and B Low RMSD values are best, zero indicates exact equality between the (sub)structures

Chapter 9 Superposition and Dynamic Programming 3 Root Mean Square Deviation Two different measures are mainly used for scoring the transformation (superposition) Coordinate RMSD Distance RMSD

Chapter 9 Superposition and Dynamic Programming 4 Coordinate RMSD Let (  1,  1 ),..., (  r,  r ) be the coordinate set (three values each) of the equivalent elements of the equivalence E The problem is then to find a transformation T for A which minimizes the coordinate root mean square deviation, that is where w i are weights corresponding to each pair (  i,  i ), and often set to one. A transformation can be performed by –A translation (three distances) –A rotation (three angles, around the x-, y-, z-axis) –The rotation can also be performed around one axis, the direction of which has to be calculated for each rotation

Chapter 9 Superposition and Dynamic Programming 5 Coordinate RMSD, cont’ A transformation for the minimum RMSD can be found by 1.Shift the centroids (geometrical centres) of each structure to the origin of a common coordinate system 2.Find the rotation of A that minimizes the RMSD C A rotation around the origin can be described by an orthogonal matrix R 3,3 (3D space) with determinant equal to 1 –There exist equations describing the connections between the angles (3) and the values of the matrix (9) A matrix is orthogonal if –The scalar product of any two different columns is 0 –The scalar product of any column with itself is 1 The deteminant of a 3 x 3 matrix R with elements {r ij } is calculated as r 11 r 22 r 33 + r 12 r 23 r 31 + r 13 r 21 r 32 - r 11 r 23 r 32 - r 12 r 21 r 33 - r 13 r 22 r 31

Chapter 9 Superposition and Dynamic Programming 6 Coordinate RMSD, cont’ The orthogonal requirement is for assuring that the distances between the points are not changed (ridig body) The determinant requirement is for not ”reflecting” (”mirrorring”) the structure

Chapter 9 Superposition and Dynamic Programming 7 Example A point (1, -1, 1) are to be rotated by an orthogonal matrix with deteminant equal to one Show that the matrix is orthogonal Show that the determinant is one Find the new coordinates

Chapter 9 Superposition and Dynamic Programming 8 Coordinate RMSD, cont’ The formula can therefore be described by a rotation matrix R and a translation vector t, and we search for a pair (R,t) which minimizes the expression (assuming w i =1) Since t is found by moving to common centroids, the problem can be formulated as finding the orthogonal matrix, with determinant one, that minimizes the function where (  1,  1 ),..., (  r,  r ) are now the coordinates after the structures are moved to common origin Algorithms exist for finding such a matrix

Chapter 9 Superposition and Dynamic Programming 9 Distance RMSD The distance score method measures how equal corresponding pairwise distances in the two structures are It alleviates the need for finding a translation and rotation of one of the structures, and is therefore faster where  ij A is the spatial distance between the elements of A in pairs i and j of the equivalence However, it has a (serious) weakness, it is invariant under reflection The two measures are experimentally shown to have a close to linear relation

Chapter 9 Superposition and Dynamic Programming 10 Using RMSD as scoring of structure similarities The problem of pairwise structure comparison is often to find equivalences with low RMSD values Several quite different equivalences with similar scores might be found, which one is ”correct” is often not easy to determine However, always consider the number of elements in the equivalences –for random comparisons the expected RMSD values seems to be proportional to the square root of the number of equivalent residues Different measures can then be used for evaluating how well two structures can be superposed

Chapter 9 Superposition and Dynamic Programming 11 Using RMSD as scoring of structure similarities

Chapter 9 Superposition and Dynamic Programming 12 Comparison methods using superpositon and dynamic programming Dynamic programming cannot be directly used for structure comparison Two different approaches are used for combining superposition and dynamic programming –Alternating superposition and dynamic programming –”Contemporary” superposition and dynamic programming

Chapter 9 Superposition and Dynamic Programming 13 Alternating superposition and dynamic programming p:=0, E o := an initial equivalence (sometimes subalignment required) iter find the transformation T p for min RMSD, using the r pairs in E p use T p on the whole structure A, giving A* calculate a scoring matrix R ij as function of the distance between residue i in A* and residue j in B find the best alignment P with use of R (use dynamic programming) p:=p+1 E p := the r pairs in P with shortest distances in R until E p =E p-1 or p<p max P now contains a found equivalence (or alignment)

Chapter 9 Superposition and Dynamic Programming 14 Alternating superposition and dynamic programming, cont’ The scoring matrix can have more components, in addition to the distance Sequence component, how similar are the amino acids? How similar are the local environments? Number of pairs in each equivalence Instead of a fix number for r, r can vary with the number of ”overlapping” pairs