Edge Orienting Dataset

[annaritz]

Maximum Edge Orientation (MEO) is an algorithm that orients the edges in a PPI network. Should we have a dataset that evaluates this? Other methods that return directed graphs could also be evaluated this way, if the original input is an undirected network.

If we do want to have a benchmark dataset evaluating these types of methods, I think the synthetic networks from pathway databases would be useful here. Many of these pathway databases have (at least a subset of) directed edges; we can use these as ground truth and check whether pathway reconstruction methods output the correct direction for directed edges.

[agitter]

In the original MEO evaluation, I thought the most valuable evaluation metric was derived from synthetic datasets from pathway databases. We looked at the longest consecutive subpaths that were oriented correctly with respect to the pathway database. That is a niche evaluation and metric.

If using real datasets, other source-target datasets may be sufficient.