Stochastic algebraic models SAMSI Transition Workshop June 18, 2009 Reinhard Laubenbacher Virginia Bioinformatics Institute and Mathematics Department.

Slides:



Advertisements
Similar presentations
Systems biology SAMSI Opening Workshop Algebraic Methods in Systems Biology and Statistics September 14, 2008 Reinhard Laubenbacher Virginia Bioinformatics.
Advertisements

Uniform algorithms for deterministic construction of efficient dictionaries Milan Ružić IT University of Copenhagen Faculty of Mathematics University of.
Polynomial dynamical systems over finite fields, with applications to modeling and simulation of biological networks. IMA Workshop on Applications of.
A Mathematical Formalism for Agent- Based Modeling 22 nd Mini-Conference on Discrete Mathematics and Algorithms Clemson University October 11, 2007 Reinhard.
Distributed Markov Chains P S Thiagarajan School of Computing, National University of Singapore Joint work with Madhavan Mukund, Sumit K Jha and Ratul.
DYNAMICS OF RANDOM BOOLEAN NETWORKS James F. Lynch Clarkson University.
Special topics in electrical and systems engineering: Systems Biology ESE Pappas Kumar Rubin Julius Halász.
Discrete models of biological networks Segunda Escuela Argentina de Matematica y Biologia Cordoba, Argentina June 29, 2007 Reinhard Laubenbacher Virginia.
CmpE 104 SOFTWARE STATISTICAL TOOLS & METHODS MEASURING & ESTIMATING SOFTWARE SIZE AND RESOURCE & SCHEDULE ESTIMATING.
Variants of Stochastic Simulation Algorithm Henry Ato Ogoe Department of Computer Science Åbo Akademi University.
Multiscale Stochastic Simulation Algorithm with Stochastic Partial Equilibrium Assumption for Chemically Reacting Systems Linda Petzold and Yang Cao University.
Gibbs Sampling Qianji Zheng Oct. 5th, 2010.
Towards Systems Biology, October 2007, Grenoble Adam Halasz Ádám Halász joint work with Agung Julius, Vijay Kumar, George Pappas Mesoscopic and Stochastic.
Artificial Intelligence Lecture 2 Dr. Bo Yuan, Professor Department of Computer Science and Engineering Shanghai Jiaotong University
Introduction to Sampling based inference and MCMC Ata Kaban School of Computer Science The University of Birmingham.
The Rate of Concentration of the stationary distribution of a Markov Chain on the Homogenous Populations. Boris Mitavskiy and Jonathan Rowe School of Computer.
Pattern Recognition and Machine Learning
Hidden Markov Model 11/28/07. Bayes Rule The posterior distribution Select k with the largest posterior distribution. Minimizes the average misclassification.
Network Statistics Gesine Reinert. Yeast protein interactions.
1. Elements of the Genetic Algorithm  Genome: A finite dynamical system model as a set of d polynomials over  2 (finite field of 2 elements)  Fitness.
Gene Regulatory Networks - the Boolean Approach Andrey Zhdanov Based on the papers by Tatsuya Akutsu et al and others.
Exploring Network Inference Models Math-in-Industry Camp & Workshop: Michael Grigsby: Cal Poly, Pomona Mustafa Kesir: Northeastern University Nancy Rodriguez:
Genome evolution: a sequence-centric approach Lecture 3: From Trees to HMMs.
6. Gene Regulatory Networks
Discrete models of biochemical networks Algebraic Biology 2007 RISC Linz, Austria July 3, 2007 Reinhard Laubenbacher Virginia Bioinformatics Institute.
. Approximate Inference Slides by Nir Friedman. When can we hope to approximate? Two situations: u Highly stochastic distributions “Far” evidence is discarded.
Gaussian Processes for Transcription Factor Protein Inference Neil D. Lawrence, Guido Sanguinetti and Magnus Rattray.
MAKING COMPLEX DEClSlONS
Genetic network inference: from co-expression clustering to reverse engineering Patrik D’haeseleer,Shoudan Liang and Roland Somogyi.
Learning Structure in Bayes Nets (Typically also learn CPTs here) Given the set of random variables (features), the space of all possible networks.
Statistical Decision Theory
Polynomial models of finite dynamical systems Reinhard Laubenbacher Virginia Bioinformatics Institute and Mathematics Department Virginia Tech.
Pisa, 11/25/2002Susanna Donatelli1 Modelling process and heterogeneous model construction Susanna Donatelli Modelling and evaluation groups.
Modeling and identification of biological networks Esa Pitkänen Seminar on Computational Systems Biology Department of Computer Science University.
Lecture 4: Metabolism Reaction system as ordinary differential equations Reaction system as stochastic process.
Stochastic modeling of molecular reaction networks Daniel Forger University of Michigan.
Using Gröbner Bases to Reconstruct Regulatory Modules in C. elegans Brandilyn Stigler Southern Methodist University.
This presentation is made available through a Creative Commons Attribution- Noncommercial license. Details of the license and permitted uses are available.
Microarrays.
Statistical Decision Theory Bayes’ theorem: For discrete events For probability density functions.
Radhamanjari Samanta *, Soumyendu Raha * and Adil I. Erzin # * Supercomputer Education and Research Centre, Indian Institute of Science, Bangalore, India.
Tracking Multiple Cells By Correspondence Resolution In A Sequential Bayesian Framework Nilanjan Ray Gang Dong Scott T. Acton C.L. Brown Department of.
CS Statistical Machine learning Lecture 24
ECE-7000: Nonlinear Dynamical Systems Overfitting and model costs Overfitting  The more free parameters a model has, the better it can be adapted.
Equilibrium transitions in stochastic evolutionary games Dresden, ECCS’07 Jacek Miękisz Institute of Applied Mathematics University of Warsaw.
The generalization of Bayes for continuous densities is that we have some density f(y|  ) where y and  are vectors of data and parameters with  being.
Seminar on random walks on graphs Lecture No. 2 Mille Gandelsman,
Wei Sun and KC Chang George Mason University March 2008 Convergence Study of Message Passing In Arbitrary Continuous Bayesian.
Introduction to Statistical Models for longitudinal network data Stochastic actor-based models Kayo Fujimoto, Ph.D.
Review of Statistical Terms Population Sample Parameter Statistic.
Learning Chaotic Dynamics from Time Series Data A Recurrent Support Vector Machine Approach Vinay Varadan.
Nonlinear differential equation model for quantification of transcriptional regulation applied to microarray data of Saccharomyces cerevisiae Vu, T. T.,
Why use landscape models?  Models allow us to generate and test hypotheses on systems Collect data, construct model based on assumptions, observe behavior.
CS Statistical Machine learning Lecture 25 Yuan (Alan) Qi Purdue CS Nov
Instructor: Spyros Reveliotis IE7201: Production & Service Systems Engineering Fall 2009 Closure.
Ch 6. Markov Random Fields 6.1 ~ 6.3 Adaptive Cooperative Systems, Martin Beckerman, Summarized by H.-W. Lim Biointelligence Laboratory, Seoul National.
Modelling Complex Systems Video 4: A simple example in a complex way.
Stochasticity and Probability. A new approach to insight Pose question and think of the answer needed to answer it. Ask: How do the data arise? What is.
Inferring Regulatory Networks from Gene Expression Data BMI/CS 776 Mark Craven April 2002.
Path Coupling And Approximate Counting
Recovering Temporally Rewiring Networks: A Model-based Approach
Xianfeng Song Lin Wang Sima Setayeshgar
1 Department of Engineering, 2 Department of Mathematics,
Hidden Markov Models Part 2: Algorithms
1 Department of Engineering, 2 Department of Mathematics,
Estimating Networks With Jumps
1 Department of Engineering, 2 Department of Mathematics,
Network Inference Chris Holmes Oxford Centre for Gene Function, &,
The Most General Markov Substitution Model on an Unrooted Tree
Estimation from Quantized Signals
Presentation transcript:

Stochastic algebraic models SAMSI Transition Workshop June 18, 2009 Reinhard Laubenbacher Virginia Bioinformatics Institute and Mathematics Department Virginia Tech

Systems biology working group activities 1.Algebraic models of biological networks 2.ODE models of biochemical reaction networks Foci: Structure dynamics Dynamics structure Experimental design

Polynomial dynamical systems Let k be a finite field and f 1, …, f n k [ x 1, …, x n ] f = ( f 1, …, f n ) : k n k n is an n-dimensional polynomial dynamical system over k. Natural generalization of Boolean networks. Fact: Every function k n k can be represented by a polynomial, so all finite dynamical systems k n k n are polynomial dynamical systems.

Parameter estimation Problem: Given experimental time course data and a partially specified model f = ( f 1, …, f n ) : k n k n, with/without information on the function structure, estimate the unspecified functions by fitting them to the data.

Parameter estimation Variables x 1, …, x n with values in k. (E.g., protein concentrations, mRNA concentrations, etc.) (s 1, t 1 ), …, (s r, t r ) state transition observations with s j k n, t j k (E.g., consecutive measurements in a time course experiment.) Network inference: Identify a function g: k n k such that f(s j )=t j.

The model space Let I be the ideal of the points s 1, …, s r, that is, I =. Let g be one particular feasible function/parameter. Then the space M of all feasible parameters is M = g + I.

Model selection In the absence of other information, choose a model which is reduced with respect to the ideal I. Laubenbacher, Stigler, J. Theor. Biol Several other methods. Contributors: E. Dimitrova, L. Garcia, A. Jarrah, M. Stillman, P. Vera-Licona

Dimitrova, Hinkelmann, Garcia, Jarrah, L., Stigler, Vera-Licona

Model selection Model selection in original method requires choice of term order Improvement: Construct a wiring diagram using information from all term orders. Dimitrova, Jarrah, L., Stigler, A Gröbner-fan based method for biochemical network modeling, ISSAC 2007

Dynamic model Dimitrova, Jarrah: Construct a probabilistic polynomial dynamical system by sampling the reduced models in all the Groebner cones, together with a probability distribution on the models derived from cone volumes. Alternative method constructed by B. Stigler.

Probabilistic Boolean networks For each variable there is a family of Boolean functions, together with a joint probability distribution. At each update, choose a random function out of this family. Shmulevich, E. Dougherty, et al. Dimitrova, Jarrah produce a probabilistic polynomial dynamical system

Update-stochastic Boolean networks Update variables sequentially, in one of two ways: At each update, choose at random a permutation, which specifies an update order. At each update, choose at random a variable that gets updated. Sequential update is more realistic biologically. See, e.g., Chaves, Albert, Sontag, J. Theor. Biol., 2005

Philosophy: Stochastic sequential update arises through random delays in the completion time of molecular processes. Consequence: Can approximate update-stochastic systems through systems with random delays, i.e., special function-stochastic systems. This approach is taken in Polynome.

General problem Study function-stochastic polynomial dynamical systems. Note: Can be viewed as a special family of Markov chains.

Why polynomial dynamical systems?

Algebraic models Most common algebraic model types in systems biology: Boolean networks, including cellular automata Logical models Petri nets

A common modeling framework 1.Boolean networks are equivalent to PDS over the field with two elements. 2.(Jarrah, L., Veliz-Cuba) There are algorithms that translate logical models and Petri nets into PDS. LM PN PDS

T cell differentiation

Stochastic systems Instead of f = (f 1, …, f n ) : k n k n consider f = ({f 1 }, …, {f n }) : k n k n, together with probability distributions on the sets {f i }. At each update, choose the ith update function from the set {f i } at random. This is a function-stochastic polynomial dynamical system.

Special case: Delay systems Let f = (f 1, …, f n ) : k n k n be a deterministic system. Let F = ({f 1, id}, …, {f n, id}), together with a probability distribution on each set. Each time id is chosen for an update, a delay occurs in that variable. What is the effect of delays on network dynamics?

An example Theorem. (Hinkelmann, Jarrah, L.) Let f be a Boolean linear system with dependency graph D. Let F be the associated delay system. Then F has periodic points if and only if D contains directed cycles (feedback loops).

Open problems Study in more generality the effects of stochastic delays on algebraic model dynamics. Can one use stochastic delay systems to efficiently simulate deterministic sequential systems? What are good simulation methods for this purpose?