This is a python re-implementation of the algorithm described in The Inferelator, by Bonneau Labs, NYU.
It provides a method for deriving genome-wide transcriptional regulatory interactions, and predicts a large portion of regulatory networks using L2 Bayesian Regression.
The Inferelator should be run using python3 with the following structure:
python3 inferelator.py [JOB_NAME]
The JOB_NAME is the name of the configuration script that should be used, and should be located in the jobs/ directory.
For example to run with the provided dream4_bbsr_low job one would use the command python3 inferelator.py dream4_bbsr_low in the root directory
NOTE: Some configuration specifications are listed in the default configuration file but have not yet been implemented.
p['inputDir'] = 'input/dream4' - Directory in which input files will be search
p['metaDataFile'] = 'meta_data.tsv' - Meta data file to be used from the specified inpurDir
p['priorsFile'] = 'gold_standard.tsv' - Priors data file to be used from the specified inpurDir
p['goldStandardFile'] = 'gold_standard.tsv' - Gold Standard data file to be used from the specified inpurDir
p['jobSeed'] = 42 - Random Seed to be used by the application
p['saveToDir'] = None - Directory in which to save outputted files
p['verbose'] = False - Option to print computed Betas at each bootstrap
p['demo'] = False - Option to use precomputed clean data (makes beta output cleaner)
p['exportCLRMatrix'] = False - Option to write computed CLR matrices to output directory
p['exportBSDR'] = False - Option to write computed bootstrap specific design and response matricies to outpur directory
Any questions concerning running the code or creating the job script can be directed to joseph.macaluso@stonybrook.edu