Code for integrative analysis of host transcriptomic/proteomic response to pathogens.
Contains two code modules:
- A. MTG-LASSO to infer protein regulators of gene expression modules
- B. Physical subnetwork inference to connect predicted regulators for modules
Written with MATLAB 2014b, SLEP v4.1, Python 2.7, bash.
The bash script, infer_protein_regs.sh, allows the user to run protein regulator inference using either MTG-LASSO or LASSO once the following are specified:
- Location of protein data file (text, tab-delimited matrix)
- Location of gene expression module files (one file per module)
- Output directory
We extract high-confidence protein regulators based on frequency across folds as well as setting a threshold for regression weight and testing that the protein is not randomly assigned that weight under permutations of the protein data. Main script for that step: get_highconf_regs.sh
Java, Python 2.7, GAMS, CPLEX, bash.
Finds all directed paths up to a given length that connect 'upstream' regulators (eg, signaling proteins or other protein-level regulators that indirectly influence gene expression) to 'downstream' regulators (eg, transcription factors that directly affect RNA levels).
Main script: gen_paths/get_paths_for_modules.sh
Source code for influenza_subnet.jar is in the src/ directory. Main class: apps/InfluenzaMain
Produces GAMS-ready files in directory gams_input/.
This step requires GAMS/CPLEX. (With some revisions, a different GAMS-compatible MIP solver could be used instead of CPLEX.)
Main script: optimize/run_modules.sh