Hi, All the latest code is in the repository now. I don't foresee any major changes anymore, so this could be a good time to update. To use "Run", you will have to add values for new parameters in your properties file, see attached file for an example. "RunXml" contains some sample code to process xml files. The javadoc documentation is here: /group/biocomp/projects/segal/java/javadoc/index.html Have fun, Tom -- Tom Michoel <http://www.psb.ugent.be/~tomic/> # LeMoNe properties file # # Data directory dataDir=/group/biocomp/projects/segal/java/test_data/Challenge_5 # # Expression data, tab delimited file in format: # - line 1: description of conditions # - column 1: gene names # - column 2: gene description expressiondata=data.txt # Data type (AFFY or RATIO) dataType=AFFY # Candidate regulators file regulatordata=tfs.csv # # Perform initial clustering? (true/false) initClust=true # # If you said false to initial clustering, give initial number of clusters # and cluster file, in format: # - each line has a gene name and cluster number # - cluster numbers start with 0 or 1, comment lines (#) are discarded #numclust= #clusterdata= # # Statistical parameters used for the Bayesian score lambda0=0.1 mu0=8.5 alpha0=0.1 beta0=0.1 # # Use Gibbs sampling or heuristic search to optimize modules (=gene clustering)? (true/false) gibbsSampleGenes=true # # If you said true to Gibbs sampling genes, set Gibbs sampler properties: # - number of runs numRunsGenes=3 # - number of steps in each run numStepsGenes=100 # - number of initial steps before sampling starts burnInGenes=10 # - number of steps between samples sampleStepGenes=100 # # If you said false to Gibbs sampling genes, set heuristic search convergence cutoff # and whether or not you want an output file with all score values during search # (convergence is measured in Bayesian score per gene) #epsConvergence=0.001 #returnScore=false # # Use Gibbs sampling or heuristic search to optimize experiment partitioning within a module? (true/false) gibbsSampleExpts=true # If you said true to Gibbs sampling experiments, set Gibbs sampler properties: # - number of runs numRunsExpts=1 # - number of steps in each run numStepsExpts=1100 # - number of initial steps before sampling starts burnInExpts=100 # - number of steps between samples sampleStepExpts=100 # # Make one consensus regulation tree out of different experiment partitions? (true/false) makeConsensusTree=true # If you said true to makeConsensusTree, give level where trees should be cut before making consensus, # minimum number of experiments in one consensus experiment cluster, and the factor to determine significant # splits cutLevel=1 minSizeExpt=15 signiFact=1.0 # # Score cutoff for pruning regulation trees (measured in Bayesian score per gene) scoregain=0.0 # # Use merge score from Bayesian Hierarchical Clustering or simple score difference # when hierarchically clustering experiments or experiment clusters? (true/false) useBHCscore=true # # Assign regulators to tree nodes stochastically or deterministically (minimum entropy)? (true/false) assignRegStoch=true # if you said true to assigning regulators stochastically, set number of regulators per node # and probability parameter beta numReg=100 betaReg=20 # Assign regulators to all levels or only down to certain level; if you say false to assignAllLevels, set the # desired level depth in assignRegLevel assignAllLevels=false assignRegLevel=2 # if you said false to assigning regulators stochastically, do you want an acyclic network? (true/false) acyclic=false # # name of XML file (without extension) to store output, # if set to 'auto' an automatic name based on date and time of run is made. XMLfilename=auto # # # Enable/disable BiNGO GO tests on modules # Be sure to have a copy of the bingo_gui.properties file # in the current directory before setting this option to "true" BiNGO=false # # Enable/disable module drawing drawModules=false