Splitpea: quantifying protein interaction network rewiring changes due to alternative splicing in cancer

This serves as the supplementary data repo for the paper, Splitpea: quantifying protein interaction network rewiring changes due to alternative splicing in cancer. 

The file structure is as follows:

Splitpea rewired PPI networks for individual patient samples (files ending in `patient-rewired-networks.zip`). The networks are stored as dat files, which are tab delimited with one row per edge. Gene ids are reported as entrez id, the edge weights determine the directionality of the interaction (positive weights are potential gains, negative weights are edges likely lost). Chaos edges can have a positive or negative weight but will be indicated by a boolean value in the `chaos` column.

The consensus networks (files ending in `consensus_networks`) are separated into two networks one for positive edges and one for negative edges present in at least 80% of tumor samples. These files are tab delimited files with one row per edge in the corresponding undirected weighted network. The following columns are included:

node1, node2: genes incident on the edge as entrez gene IDs

mean_weight: average weight of edge across networks (where edge is present) used to build the consensus

total_weight: sum of weights across networks (where edge is present) used to build the consensus

num_graphs: # of networks where edge was present

prop_graphs: # of networks where edge was present / total # of networks used to build the consensus

IRIS.zip: Splicing matrix spliced exon files from IRIS (Pan et al. https://doi.org/10.1073/pnas.2221116120). These are the initial inputs for the Splitpea run described in the paper.

BRCA-psi.zip: precalculated PSI values for breast cancer samples as described in the paper

PAAD-psi.zip: precalculated PSI values for pancreatic cancer samples as described in the paper

*-centralities.zip: precalculated centralities for each patient network

*-pickle.zip: network representations as pickle files