# Cohort data ------------------------------------------------------------------

Cohorts data - mRNA / slides and response information:

* R lists of bulk mRNA datasets ................................................
  
    - ./data/all_mRNA_dedup_HRpos.rds -> ER+/HER2- chemotherapy datasets 
    - ./data/all_mRNA_dedup_HER2pos.rds -> HER2+ anti-HER2 datasets 
    - ./data/all_mRNA_dedup_TNBC_w_brightness.rds -> TNBC chemotherapy datasets  
    - ./data/all_mRNA_dedup_HRpos_ICB.rds -> exploratory ER+/HER2- ICB datasets 

* R lists of response information (same structure as mRNA datasets) ............

    - ./data/all_resp_HRpos.rds  -> ER+/HER2- chemotherapy datasets 
    - ./data/all_resp_HER2pos.rds -> HER2+ anti-HER2 datasets 
    - ./data/all_resp_TNBC_w_brightness.rds -> TNBC chemotherapy datasets  
    - ./data/all_resp_HRpos_ICB.rds -> exploratory ER+/HER2- ICB datasets 

* R lists of slides-inferred mRNA with path2Omics and responses ................

    - ./data/slides_data/all_mRNA_slides.rds
    - ./data/slides_data/all_responses_slides.rds

# For supplementary performances and additional results you can check:

    - All files starting with supp_table... in ./data
    - All .rds objects in ./data/performances 

# How to run the code ----------------------------------------------------------

NOTE: Two datasets have been removed prior to publication due to data-sharing
restrictions:

  - phs001291 (dbGaP): part of the HER2+ anti-HER2 therapy cohort. Overall
    performances within this subtype may differ slightly from the paper.

  - pusztaiICB (controlled-access): part of the TNBC ICB exploratory cohort.
    Exploratory ICB analyses for TNBC are based on GSE194040_pembro and
    GSE173839 only.

  - phs002419 (dbGaP): part of the ER+/HER2- ICB exploratory cohort.
    Exploratory ICB analyses for ER+/HER2- are based on GSE194040_pembro and
    GSE173839 only.


The code is divided in two blocks:


* ```run_all_01.sh``` will run all the code of expression-based analyses 

    01 - load expression data, deconvolve and fit BRIDGE models on training data. 
    02 - compute the BRIDGE response scores on all the testing datasets 
    03 - computes overall performances on testing data (AUCs, ORs, .... And plots)
    04 - computes performances and plots running PAM50-correlation based model 
    05 - computes performances and plots quintiles analysis and single abundances. 

    You may run the code:

    ```bash run_all_01.sh step1``` if you want to run just one specific step or 
    ```bash run_all_01.sh all``` if you want to run everything sequentially 



* ```run_all_02.sh``` will run all the code of slides-based analyses 
     
    01 - load path2Omics processed data and runs BRIDGE. 
    02 - computes performances of slides based predictions  


    You may run the code:

    ```bash run_all_02.sh step1``` if you want to run just one specific step or 
    ```bash run_all_02.sh all``` if you want to run everything sequentially 

