• Corpus ID: 3344038

Bayesian variable selection in linear dynamical systems

  title={Bayesian variable selection in linear dynamical systems},
  author={Atte Aalto and Jorge M. Gonçalves},
  journal={arXiv: Methodology},
We develop a method for reconstructing regulatory interconnection networks between variables evolving according to a linear dynamical system. The work is motivated by the problem of gene regulatory network inference, that is, finding causal effects between genes from gene expression time series data. In biological applications, the typical problem is that the sampling frequency is low, and consequentially the system identification problem is ill-posed. The low sampling frequency also makes it… 

Figures and Tables from this paper

Continuous time Gaussian process dynamical models in gene regulatory network inference

A GRN inference method called BINGO is developed, based on MCMC sampling of trajectories of the GPDM and estimating the hyperparameters of the covariance function of the Gaussian process, which is superior in dealing with poor time resolution and computationally feasible.

Linear system identification from ensemble snapshot observations

Two different paradigms are studied for linear system identification, based on tracking the evolution of the distribution of cells over time and the so-called pseudotime concept, identifying a common trajectory through the state space, along which cells propagate with different rates.

Network Stability, Realisation and Random Model Generation

This work provides procedures to obtain "stable" DSF models or require the presence of feedback structures while keeping topology and dynamics random up to these constraints and suggests model generation algorithms, whose implementations are now publicly available.

Gene regulatory network inference from sparsely sampled noisy data

BINGO is presented, a powerful method for network inference from time series data that clearly and consistently outperforms state-of-the-art methods and is available to any researcher, helping to decipher the complex mechanisms of life.

A state space representation of VAR models with sparse learning for dynamic gene networks.

A new calculation technique for EM algorithm that does not require the calculation of inverse matrices is introduced, which is applied to time course microarray data of lung cells treated by stimulating EGF receptors and dosing an anticancer drug, Gefitinib.

Identification of Sparse Continuous-Time Linear Systems with Low Sampling Rate: Exploring Matrix Logarithms

This paper considers linear noise-driven dynamical systems evolving in continuous time and provides theoretical results for when a unique solution exists up to a finite equivalence class and a mixed integer linear programming formulation corresponding to a simplified version of the problem.

Model selection for dynamical systems via sparse regression and information criteria

An algorithm for model selection is developed which allows for the consideration of a combinatorially large number of candidate models governing a dynamical system, and it is shown that AIC scores place each candidate model in the strong support, weak support or no support category.

A Bayesian Lasso via reversible-jump MCMC

Bayesian Variable Selection in Linear Regression

Abstract This article is concerned with the selection of subsets of predictor variables in a linear regression model for the prediction of a dependent variable. It is based on a Bayesian approach,

Reversible jump Markov chain Monte Carlo computation and Bayesian model determination

Markov chain Monte Carlo methods for Bayesian computation have until recently been restricted to problems where the joint distribution of all variables has a density with respect to some fixed

MCMC methods for diffusion bridges

The novel algorithmic idea of the paper is that proposed moves for the MCMC algorithm are determined by discretising the SPDEs in the time direction using an implicit scheme, parametrised by θ ∈ [0,1].

MCMC Methods for Functions: ModifyingOld Algorithms to Make Them Faster

An approach to modifying a whole range of MCMC methods, applicable whenever the target measure has density with respect to a Gaussian process or Gaussian random field reference measure, which ensures that their speed of convergence is robust under mesh refinement.


This paper describes and compares various hierarchical mixture prior formulations of variable selection uncertainty in normal linear regression models. These include the nonconjugate SSVS formulation