READ ME FOR "Sign-based shrinkage based on an asymmetric LASSO penalty"
Manuscript ID: JDS2103-006

Contact Info: Eric S. Kawaguchi (eric.kawaguchi@med.usc.edu)

Code to reproduce simulation results are located in this folder.

Before running the R scripts please download the following R package from Github:
https://github.com/erickawaguchi/xrnet/tree/esk_development

An R script to do this is provided in the directory as well: preamble.R

Load the R project (asymLASSO.Rproj) since it will automatically set the working directory.

#######################################################################################
# This directory holds the following:

1. simulationList.xlsx: Holds all simulation parameters that were performed in the main manuscript (and supplemental material).
	a. Sheet 1 (model selection - low): Parameters for low-dimensional OLS and logistic regression.
	b. Sheet 2 (model selection - high): Parameters for high-dimensional OLS regression.
	c. Sheet 3 (sign): Parameters for evaluating switched design matrix (sign invariance).
	d. Sheet 4 (tau): Parameters for evaluating sensitivity to skewness parameter tau.
 
2. results/: Folder that contains ALL simulation results. The R scripts to perform other simulation studies are outputted as .RData files and are housed in this directory. Some .RData files to reproduce the tables in the main text are included in this folder.

3. sourceFiles/: Folder that contains additional R scripts to compute correlation matrices, measures for evaluation, etc. This folder is typically sourced before the simulations are performed.

#######################################################################################
The following R scripts are as follows

Running the R scripts AS IS (no modifications) will create .RData files (stored in the results directory) that house the results found in the MAIN text. 

1. sim-s1-tau-xx.R: Performs simulation studies for Section 3.1. The xx inform which correlation structure you are interested in running (ar = autoregressive, cs = compound symmetry/equicorrelation, ind = independent). The main text uses the -ind suffix (Table 1). The -ar and -cs suffixes can be found in Tables S1 and S2 in the Online Supplemental Material, respectively. 

2. sim-s2-OLS-xx.R: Performs simulation studies for Section 3.2 (Tables 2, 3, 4). The xx informs which correlation structure to run. The main text uses the -ar suffix (Tables 2 and 3). The -ind and -cs suffixes are used in the supplemental material (Table S3). sim-s2-HD-OLS refers to the high-dimensional simulation (Table 4) in Section 3.2. It is identical to sim-s2-OLS except that the Excel sheet to extract simulation parameters and the effect estimates are modified in the code. 

3. sim-s3-sign-xx.R: Performs simulation studies for Section 3.3. Again, xx informs which correlation structure you are interested in running. The main text uses the -ind suffix. The -ar and -cs suffixes are used in the supplemental material (Figures S1 and S2, respectively).

4. sim-s4-logistic.R: Performs simulation studies for Section 3.4. We ONLY considered the AR correlation structure, thus no suffix is included. This corresponds to Table 5 in the main text.

5. evaluation-tau.R: R script to get the results table (in LaTex format) for tau simulation (Section 3.1)

6. evaluation-modsel.R: R script to get the results table (in LaTex format) for model selection simulation (Section 3.2 & Section 3.4)

7. evaluation-sign.R: R script to get the results table (in LaTex format) for sign invariance simulation (Section 3.3)

8. RDA_TCGA.R: Runs the real data analysis.
