Processing in Python

These files provide the steps for extracting informative features of each modality, generating a network per modality and performing Similarity Network Fusion (SNF).

Prior to performing processing in python, each modality is required to be saved in the format {modality}_preprocessed.pkl. See GettingStarted for more.

Step 1 - Preprocessing

Create a folder called for the dataset called data and within this folder copy the preprocessed data into a folder name e.g. TCGA-BRCA.

Run Preprocessing.ipynb for each modality of interest. There are two options for the feature extraction at this stage. Either differential gene expression or elastic net regression.

Save each modalities processed folder with naming convention {modality}_processed.pkl in the data/raw Save each network with naming convention {modality}_graph.graphml in the data/Network folder

The options are
BRCA :
target = ‘paper_paper_BRCA_Subtype_PAM50’

LGG :
target = ‘paper_Grade’

KIPAN :
target = ‘subtype’

Step 2 - Similarity Network Fusion

Create a folder called Network outside data \

Specify the modalities of interest in the list modalities

Point the SNF script to the new Network folder

Run the cell to perofrm SNF for the modalities of interest.

Example of directory structure for TCGA-BRCA

• data

  • TCGA-BRCA

    • mRNA

      • mRNA.pkl

    • miRNA

      • miRNA.pkl

  • raw

    • mRNA_processed.pkl

    • miRNA_processed.pkl

  • Network

    • mRNA_graph.graphml

    • miRNA_graph.graphml

    • mRNA_miRNA_graph.graphml