Elastic net regression training and testing (LENP)

For training and testing LENP models:

Input:

Expression matrix: Expression/filtered_lncRNA_z_EXP.csv

Drug responses data: drug_res/drug_res_merge.csv

Operation: (*lead to one or multi output)

1. Parameter Optimization*

Output: optimized parameter for 265 drugs (IC50 and AUC)

2. Bootstrapping

Using optimized parameters, compute the predictive scores for each drug (IC50 and AUC)

3. Get top predictive genes*

Sort the genes by predictive scores and get top 20 genes.

Output: predictive score of all genes for each drug (IC50 and AUC)

4. Final model construction*:

i. Get coefficients

Output: Average coefficiencts of every predictors in each model (IC50 and AUC)

ii. Model performance assessments

Pearson's correlation coefficients, kendall's tau