This repository contains code for processing kinobeads and kinomescan data for cell line response prediction, as outlined in the paper: [https://www.biorxiv.org/content/10.1101/2022.12.06.519165v1]
This repository contains code written almost entirely in R, using Rstudio and 'Tidyverse' idioms. The file package_check.R describes all the R packages needed to run the code in a covenient "pacman" script. Running this script will install all the required packages in one go.
This repository is divided into three main folders:
src: source code for generating all results and figuresdata: raw data used by the source code (included in zenodo)results: results generated by source code (not included here)figures: figures generated by source code
The folder src/data_organization contains code to process kinome profiling data from kinobeads and KINOMEscan assays, and link it to cell line responses and baseline multi-omics data.
The folder src/LINCS_modeling contains code to build machine learning models using the combined dataset, predicting outcomes of IC50 and AUC. This also includes code to process experimental data and validate model predictions.
The majority of figures included in the paper are produced directly in the model-building and analysis code, however the code for some specific visualizations can be found in the folder src/LINCS_modeling/figure_building
All the figures published in the paper can be found in the folder figures and some specific figures generated as part of model building code can be found in figures/PRISM_LINCS_klaeger