MEXCOWalk: Mutual Exclusion and Coverage Based Random Walk to Identify Cancer Modules

This is the original repository for the MEXCOwalk paper

Installing the dependencies

pip install -r requirements.txt

Input

Graph Construction

There are two files for the graph construction: the Protein Protein Interaction (PPI) network edges file and the nodes to index mapping file. Both files are located in data folder

1. The PPI network file:

The file is located at data/hint_edge_file.txt

node_i_Index node_j_Index weight
0 1 1
0 2 1

2. The nodes to index file mapping:

The file is located at data/hint_index_file.txt

index GeneName
1 A1BG
2 A1CF

Mutation Data

The files needed for calculating the mutex and coverage scores are:

1. Gene and the list of patients where that gene is mutated: The file is located at data/genes_vs_patients_indices_gen_paran.txt

RNF14	109	241	...
UBE2Q1	581	837	...
....

The patients are referred to with their IDs.

2. The patients name to patient ID mapping:

The file is located at data/patients_to_indices_gen.txt

TCGA-06-0649	1
TCGA-BA-5149	2
TCGA-DK-A1A6	3

The patient IDs start from 1

3. The gene and its corresponding mutation frequency:

This file contains the mutation frequncies, which are assigned as heats during the random walk.

The file is located at data/pan12gene2freq.txt. the file was downloaded from here

A1BG 0.00353697749196
A2M 0.0128617363344
A4GALT 0.00064308681672

Cosmic genes

The cosmic genes are loaded from Census_allTue_May_23_12-08-15_2017.tsv file, which is located in data folder. This file is used to find genes with cosmic overlap i.e. the genes in the found modules that are also present in the cosmic database.

Run

There are two bash scripts, one for running the modules discovery algorithm, the other for evaluation.

For more details on the execution parameters please refer to the bash files.

1. Modules discovery:

sh execute_all.sh

2. Evaluation

sh evaluate_all.sh

Outputs

The 'execute_all.sh' file will output modules of genes in the 'connected_components_isolarge/mexcowalk' folder.

The 'evaluate_all.sh' file will output cosmic overlap, number of modules, average and total counts of mutex, coverage and density as well as weighted average values of mutex, inverse coverage and density. The files can be found in 'out/evaluation_tab'

The evaluation file can also be run with multiple models and compared to MEXCOwalk. Simply store the models in 'out/connected_components_isolarge/model_x' and add the model name to the evaluations section in the evaluation file as follows:

-m mexcowalk model_1 model_2 model_3

Knearest

CoxPh analysis