Specifying input and output formats

[agitter]

We can finalize the universal input and output formats that we have previously discussed. Once these are locked in, we can document them in the main readme.

Input file formats:

• How to list interactions
• How to optionally denote edge directionality (or other edge features?)
• How to provide node annotations

Output file format:

• How to list interactions and specify directionality (required?)
• How to provide node annotations (is this needed?)

[annaritz]

Picking up on this issue related to input file formats. @prameshsingh has added some in the example/ directory in the input-data branch. We propose:

1. Adding a README to this directory (suggested by @agitter)
2. Having two file types: one that denotes a graph (a two-column edgelist, with optional additional cols), and one that denotes a node list (a one-column list of node names, with optional additional cols). For example, if we are given EGFR1-sources.txt and EGFR1-targets.txt node lists for EGFR, then EGFR1-nodes.txt will be generated from these files for PathLinker in prepareInputs().
3. Write code in PathLinker's prepareInputs() to actually generate this file.

For now, we're leaving the large interactome and the *-edges.txt files as-is.

[agitter]

At our recent meeting, we decided to add a header row to the SPRAS pathway output file format. This will disambiguate the third column of numeric scores, which is currently a rank for individual pathways but can also be an edge frequency in an ensemble pathway.

[agitter]

Closed by #142, which added a markdown file documenting the output file format.