Portability of cancer effects across genetic ancestries

This github hosts the code for the publication: insert link

Reporting summary

Additional information for the reporting can be found in the directory paper.

paper/ 
├── reporting_summary.txt
│
├── summary_subset_interaction_effect/ 
│   └── summary_sample_n.csv             # Number of samples per comaprison
│
├── summary_subset_prediction_effect/
│   └── summary_sample_n.csv  
│
└── summary_limma_interaction_effect/
    └── summary_sample_n.csv

Same samples are used across all analysis. For subsetting analysis (subset_interaction_effect_script.R & subset_prediction_effect_script.R) the same subsets (same patient ids) are reused!

Data download

The molecular data to run this script is downloaded from GDC Data Portal using the GDC Data Transfer Tool with the manifest.txt, metadata.json and the corresponding config_<molecular_level>.json files in configs/data/.

configs/ 
└── data/ 
    ├── TCGA_BRCA/ 
    │    ├── GDC_manifest_TCGA_BRCA_meth.txt     
    │    ├── GDC_manifest_TCGA_BRCA_mrna.txt     
    │    ├── GDC_metadata_TCGA_BRCA_meth.json    
    │    ├── GDC_metadata_TCGA_BRCA_mrna.json    
    │    ├── config_meth.json            # Config for script `gdc_process_meth.R`
    │    ├── config_mrna.json            # Config for script `gdc_process_mrna.R`
    │    └── metadata_map.csv            # Map to rename column names and values
    │
    ├── TCGA_THCA/ 
    │    └── ...     
    │
    └── TCGA_UCEC/ 
         └── ...

Once you have installed the gdc-client you can run the scripts gdc_process_meth.R, gdc_process_meth.R. Please be aware that the scripts ask for the location of the client. The scripts will download the molecular data (from GDC Portal), additional clinical data (cBioportal) and genetic ancestry (https://doi.org/10.1016/j.ccell.2020.04.012). Once the data for each molecular layer (methylation & expression) is downloaded the data is combined into one object using gdc_create_study.R. The scripts will create a new directory download/.

download/ 
└── GDCportal/ 
    ├── TCGA_BRCA/ 
    │    ├── meth_files/                 # Raw patient files
    │    ├── mrna_files/     
    │    ├── demo_meth.png               # Demographic plots
    │    ├── demo_mrna.png       
    │    ├── full_matrix_meth.rds        # All downloaded features                    
    │    ├── full_matrix_mrna.rds
    │    ├── gene_level_matrix_meth.rds  # Summarized on gene level
    │    ├── gene_level_matrix_mrna.rds    
    │    ├── metadata_meth.rds           # Metadata files
    │    ├── metadata_mrna.rds
    │    ├── probe2gene.csv              # Mapping of probe to gene
    │    └── TCGA_BRCA_omics_layers.rds  # Combined study object (meth & mrna)
    │
    ├── TCGA_THCA/ 
    │    └── ...
    │
    └── TCGA_UCEC/ 
         └── ...

Data simulation

Both data simulations (synthetic and permuted) were performed using data from TCGA BRCA. The configs for these analysis are in configs/sims/. The scripts sim_synthetic_script.R and sim_permutation_script.R will use the same config files and create results/sims/analysis. The scripts sim_synthetic_figures.R and sim_permutation_figures.R will create the figures for the publication.

results/ 
└── sims/ 
    ├── analysis/ 
    │    ├── TCGA_BRCA/
    │    │    ├── permutation_sim/
    │    │    │    ├── res_file1            # One res file per config
    │    │    │    └── ...
    │    │    │
    │    │    └── synthetic_sim/
    │    │         ├── res_file1            # One res file per config
    │    │         └── ...
    │    │
    │    └── TCGA_UCEC/ 
    │         └── ...
    │
    ├── figures/ 
    └── tables/

Differential expression and ML prediction

Both analysis differntial expression and ML prediction use the same config files in conigs/tcga/. The scripts subset_interaction_effect_script.R and subset_prediction_effect_script.R will use the same config files and create results/tcga/analysis. The scripts subset_interaction_effect_figures.R and subset_prediction_effect_figures.Rwill create the figures for the publication.

results/ 
└── tcga/ 
    ├── analysis/ 
    │    ├── TCGA_BRCA/
    │    │    ├── subset_interaction_effect/
    │    │    │    ├── res_file1            # One res file per config
    │    │    │    └── ...
    │    │    │
    │    │    └── subset_prediction_effect/
    │    │         ├── res_file1            # One res file per config
    │    │         └── ...
    │    │
    │    ├── TCGA_THCA/ 
    │    │    └── ...
    │    │
    │    └── TCGA_UCEC/ 
    │          └── ...
    │
    ├── figures/ 
    └── tables/

Requirements

Before running these scripts, please install the CrossAncestryGenPhen package from GitHub.

Link: https://github.com/DKatzlberger/CrossAncestryGenPhen

You can do this in R using:

install.packages("devtools")  # if not already installed
devtools::install_github("DKatzlberger/CrossAncestryGenPhen")

Make sure the installation completes successfully, then load the package with:

library(CrossAncestryGenPhen)

Once the package is installed and loaded, the scripts should run as intended.