GPSIM software Version 0.1211 Tuesday 19 May 2009 at 11:09 Version 0.1211 -------------- Minor bug fix in demBarenco1.m Version 0.121 ------------- Minor fixes of the code for dealing with white noise handling in multiKern. Version 0.12 ------------ Release with first draft of journal paper. Version 0.111 ------------- Mistakenly the technical noise was not being added in the kernel computation, change to gpsimUpdateKernels.m. Version 0.11 ------------ Release with scripts to recreate results in NIPS paper and at NIPS workshop talk. Updated release with correction to Hessian for non-linear response model and modular code for the MAP approximation. Removed update of kernel parameters 'Hessian correction'. Version 0.1 ----------- A quick and dirty release for the results to be submitted to the GPIP workshop. MATLAB Files ------------ Matlab files associated with the toolbox are: gpsimXGradient.m: ... distfit.m: Fit a distribution to given parameter percentiles gpsimLoadBarencoTestData.m: Load in Martino Barenco's test data as processed by mmgMOS. demBarencoVariational1.m: Optimise model using variational approximation with RBF kernel and exponential response. gpsimMapEcoliResults.m: Plot the results from the MAP script. gpsimLoadBarencoPUMAData.m: Load in Martino Barenco's data as re-processed by mmgMOS. demBarencoRank.m: Do ranking experiments on data from Barenco et al in Genome Biology. cgpsimExpandParam.m: Expand params into model structure. gpsimToolboxes.m: Toolboxes for the GPSIM software. gpsimTest.m: Test the gradients of the GPSIM model. gpsimMapUpdateYpred.m: Update the stored numerical prediction for y. gpdisimExtractParam.m: Extract the parameters of a GPDISIM model. gpsimMapFunctionalLogLikelihood.m: Compute the log likelihood of a GPSIMMAP model. gpdisimLogLikelihood.m: Compute the log likelihood of a GPDISIM model. gpsimModelFunctions.m: Update the nonlinear transformation of f. gpsimMapUpdatePosteriorCovariance.m: update the posterior covariance of f. gpsimArtificialProtein.m: Generate an artifical protein sequence. demEcoliMap1.m: Optimise model using MAP approximation with MLP kernel and multiple repression response, using SCG optimisation and Ecoli data set. gpsimAddCandidate.m: Add candidate genes to a GPSIM model. demBarencoMap1.m: Optimise model using MAP approximation with MLP kernel and MM activation response, using SCG optimisation and new PUMA processed data. gpsimCandidateLogLikeGradients.m: Compute the gradients for the parameters of candidate genes. expvarMeanOut.m: Output of an EXPVARMEAN model. cgpdisimExpandParam.m: Expand params into model structure. gpsimMapWGradient.m: Compute the gradients of W with respect to the parameters of the k-th gene. demToyProblem3.m: Generate an artifical data set and solve with GPSIM. gpdisimLogLikeGradients.m: Compute the gradients of the log likelihood of a GPDISIM model. gpsimMapGradients.m: Compute the gradients of the log likelihood of a GPSIMMAP model. gpsimMapUpdateG.m: Update the nonlinear transformation of f. gpsimMapObjective.m: Compute the objective function of the GPSIMMAP model. gpsimSampleTest.m: Test the single input motif code. gpsimMapUpdateYpredVar.m: Update the variance for y. gpsimMapFunctionalWGradient.m: computes the implicit part of the gradient gpsimCandidateLogLikelihood.m: Compute the log likelihood of a gene. gpsimSampleHMC.m: Do HMC sampling for the GPSIM model. gpsimMapUpdateF.m: Update posterior mean of f. cgpsimLogLikeGradients.m: Compound GPSIM model's gradients. demBarencoMap3.m: Optimise model using MAP approximation with MLP kernel and linear response, using SCG optimisation and new PUMA processed data with noise variance estimation. gpsimLogLikeGradients.m: Compute the gradients of the log likelihood of a GPSIM model. gpsimMapUpdateKernels.m: Updates the kernel representations in the GPSIMMAP structure. gpdisimOptimise.m: Optimise the GPSIM model. drosPlotExpPcts.m: Plot expression values gpsimMapTest.m: Tests the gradients of the GPSIMMAP model. gpsimCandidateCovGrads.m: Sparse objective function gradients wrt Covariance function. gpsimDisplay.m: Display a Gaussian process model. gpsimMapExtractParam.m: Extract the parameters of a GPSIMMAP model. gpsimMapFunctionalExtractParam.m: Extract the function values from a GPSIMMAP model. gpsimMapFunctionalUpdateW.m: Update the data component of the Hessian. gpsimCandidateOptimise.m: Optimise the GPSIM model for candidate genes. gpsimLoadBarencoData.m: Load in Martino Barenco's data as processed by mmgMOS. gpsimArtificialGenes.m: Give artifical genes from given parameters. gpsimCandidateExpandParam.m: Expand the given parameters for a candidate gene. demBarenco1.m: Run experiments on data from Barenco et al in Genome Biology. demGpdisimMef2.m: Run experiments on Mef2 data. The raw data is pre-processed by the PUMA package. demBarencoMap2.m: Optimise model using MAP approximation with MLP kernel and exp response, using SCG optimisation and new PUMA processed data. expvarMeanCreate.m: Creates an the mean function for the EXP kernel. gpsimMapLogLikelihood.m: Compute the log likelihood of a GPSIMMAP model. cgpsimLogLikelihood.m: Compound GPSIM model's log likelihood. cgpdisimLogLikelihood.m: Compound GPDISIM model's log likelihood. gpsimMapLogLikeGradients.m: Compute the gradients of the log likelihood of a GPSIMMAP model. demBarenco2.m: Run experiments on data from Barenco et al in Genome Biology. gpsimMapMarginalLikeliGradient.m: Compute the gradients of the log marginal likelihood of a GPSIMMAP model with respect to the model parameters. cgpsimExtractParam.m: Extract parameters from compound GPSIM model. gpsimMapGradFuncWrapper.m: wraps the log-likelihood function and the gradient function cgpdisimLogLikeGradients.m: Compound GPDISIM model's gradients. demToyProblem1.m: Generate an artifical data set and solve with GPSIM. gpsimCreate.m: Create a GPSIM model. gpsimOptimise.m: Optimise the GPSIM model. gpsimGradient.m: Gradient wrapper for a GPSIM model. expvarMeanExpandParam.m: Update expvarMean mapping with new vector of parameters. gpsimMapInitParam.m: Creates a set of options for a GPSIMMAP model as the initial parameters. expvarMeanExtractParam.m: Extract weights and biases from an EXPVARMEAN mapping. demToyProblem2.m: Display results from toy problem point at a time. gpdisimGradient.m: Gradient wrapper for a GPDISIM model. gpdisimExpandParam.m: Expand the given parameters into a GPDISIM structure. demToyProblem5.m: Generate an artifical data set and solve with GPSIM. drosFindTargets.m: returns a sorted list of targets drosGetGeneinds.m: returns indices of given genes in the given data gpsimMapFunctionalLogLikeGradients.m: Compute the functional gradient for GPSIMMAP. gpsimMapBarencoResults.m: Plot the results from the MAP script. gpsimMapFunctionalLogLikeHessian.m: Compute the functional Hessian for GPSIMMAP. gpdisimCreate.m: Create a GPDISIM model. gpsimCandidateObjective.m: Wrapper function for GPSIM candidate gene objective. gpsimLoadEcoliFullData.m: Load in E. coli full data for the repression case. gpsimMapCreate.m: Create a GPSIMMAP model. gpsimUpdateKernels.m: Updates the kernel representations in the GPSIM structure. gpsimObjective.m: Wrapper function for GPSIM objective. gpsimMapLikeGradientImplicit.m: computes the implicit part of the gradient gpsimLoadBarencoMASData.m: Load in Martino Barenco's data as processed by MAS5. cgpdisimExtractParam.m: Extract parameters from compound GPDISIM model. gpsimExpandParam.m: Expand the given parameters into a GPSIM structure. gpsimExtractParam.m: Extract the parameters of a GPSIM model. gpdisimGetDrosData.m: Get Drosophila data as processed by mmgMOS. gpsimCandidateUpdateKernels.m: Updates the kernel representations in the GPSIM candidate structure. gpsimOptions.m: Creates a set of default options for a GPSIM model. gpsimLoadEcoliData.m: Load in E. coli data for the represion case. gpsimMapFunctionalExpandParam.m: Expand the given function values into a GPSIMMAP structure. gpdisimObjective.m: Wrapper function for GPDISIM objective. gpsimMapExpandParam.m: Expand the given parameters into a GPSIMMAP structure. gpsimLogLikelihood.m: Compute the log likelihood of a GPSIM model. gpsimCandidateExtractParam.m: Extract the parameters of a GPSIM model. gpsimMapOptions.m: Creates a set of default options for a GPSIMMAP model. gpsimMapFunctionalLikeGrad2.m: Compute the functional gradient for GPSIMMAP. drosFindSIMTargets.m: returns a list of targets only controlled by given tf