R data and code for the paper: "A Bayesian Nonparametric Approach for Inferring Drug Combination Effects on Mental Health in People with HIV".
The data analyzed in Section 4 “Application: WIHS Data Analysis” of the paper are from the Women's Interagency HIV Study (WIHS), which is a multisite, longitudinal cohort study of the natural and treated history of women living with HIV and women at-risk for HIV in the United States. The data are publicly available. However, one need to fill in a request form for access. Full details of the data are available at https://statepi.jhsph.edu/wihs/wordpress/. R data for the simulation study and the WIHS data analysis of the paper are available at the following link: https://drive.google.com/file/d/1U2lZK9UwS60ABzhO5AtOV7ohPsnG49l1/view?usp=sharing.
The R scripts in the folder “BNP_DrugComb_Simulation” are for Section 3 “Simulation Study”, and the R scripts in the folder “BNP_DrugComb_WIHS” are for Section 4 “Application: WIHS Data Analysis”.
Libraries and Version Numbers: R 3.5.1, Rcpp 1.0.0, RcppArmadillo 0.9.700.2.0, RcppEigen 0.3.3.5.0, Matrix 1.2-14, LaplacesDemon 16.1.1, ggplot2 3.1.0, gridExtra 2.3, lattice 0.20-35.
In the folder “BNP_DrugComb_Simulation”:
-
The R script “Simulation_Main.R” reproduces Figure 4 in the manuscript, and Figure/Table S1-S9 and S14 in the Supplementary Material;
-
The R data file “Simu.Data.Preprocess.Rdata” contains the preprocessed data from the WIHS dataset used for generating simulation truths. The R script “Data_Generate.R” generates the simulated dataset, which is saved in the R data file “Simu.Truths.Rdata”;
-
The R script “MCMC_R_Functions.R” provides R functions used for MCMC, the Rcpp script “MCMC_Rcpp_Functions.cpp” provides Rcpp functions used for MCMC, and the R script “BNP_DrugComb_MCMC.R” provides the MCMC main function;
-
We save the MCMC posterior samples for one randomly selected simulated dataset in the simulation study in the R data file “Simu.MCMC.Results.Rdata", and the MCMC posterior samples for the same simulated dataset in the sensitivity analysis in the R data file “Sensi.MCMC.Results.Rdata";
-
We save the posterior co-clustering results based on 100 repeated simulations in the R data file “Simu.Co.Clustering.Rdata”, the posterior estimated number of clusters for all the 100 repeated simulations in the R data file "Simu.Cluster.Number.Rdata", and the mean squared errors for all 100 repeated simulations in the R data file “Simu.Mean.Squared.Error.Rdata”;
-
We save the estimated combination effects and the MSE in the 100 repeated simulations under our proposed method ddCRP+ST, and four alternative methods Normal+ST, Normal+Linear, DP+ST, and DP+Linear in the R data file “Simu.Combination.Effects.Rdata”.
In the folder “BNP_DrugComb_WIHS”:
-
The R script “BNP_DrugComb_WIHS_Inference.R” reproduces Figure 5-7, Table 1 in the manuscript, and Table S10 in the Supplementary Material;
-
The R script “MCMC_R_Functions.R” provides R functions used for MCMC, the Rcpp script “MCMC_Rcpp_Functions.cpp” provides Rcpp functions used for MCMC, and the R script “BNP_DrugComb_MCMC.R” provides the MCMC main function;
-
The R script “Subset_Tree_Kernel_Similarity.R” provides functions for calculating the similarity matrix induced by subset-tree kernel, and the R script “Prediction.R” provides functions for predictions；
-
We save the preprocessed data from the WIHS dataset for inference in the R data file “WIHS.Data.Preprocess.Rdata”, and the MCMC posterior samples for the WIHS data analysis under our proposed method ddCRP+ST and four altervative methods Normal+ST, Normal+Linear, DP+ST, and DP+Linear in R data files “WIHS.MCMC.Results.ddCRP.ST.Rdata”, “WIHS.MCMC.Results.Normal.ST.Rdata”, “WIHS.MCMC.Results.Normal.Linear.Rdata”, “WIHS.MCMC.Results.DP.ST.Rdata”, and “WIHS.MCMC.Results.DP.Linear.Rdata", respectively.
To facilitate the implementation of the proposed method in the decision process of HIV clinicians, and for broad application in personalized medicine, we have created an interactive web application to illustrate an example using R package shiny, available at https://wjin.shinyapps.io/Rshiny/. The web user interface interactively displays the predictive depression scores of an individual in response to the user’s choice of the individual’s clinical characteristics and ART medication use.