Add phuEGO

[agitter]

phuEGO: A network-based method to reconstruct active signalling pathways from phosphoproteomics datasets https://doi.org/10.1101/2023.08.07.552249

Available as a Python package https://github.com/haoqichen20/phuego and https://pypi.org/project/phuego/

[zhiqianx]

Notes about phuEGO

• Need to download datasets (~7GB) took 2 hrs
• Full run takes > 1hr = 1000 propagations
• Test run takes ~3 min = 10 propagations
• Seems only taking protein list from users -> message from developer: the feature using own network will be added but not sure when
• Output multiple files & networks
• Out put graphml file which visualize the pathway, require yED to open (can also set to other formats) -> maybe can help visualize SPRAS?

What does it do?

• network-based method to reconstruct active signaling pathways from phosphoproteomics datasets.
• combines three-layer network propagation with ego network decomposition to provide small networks comprising active functional signaling modules

Input data

• protein list = .txt file where first column are uniprot Ids and second columns are values associated to the importance of the protein in the first column

Output data

• numerous files generated at each steps of the algorithm -> does not really fit with SPRAS

  1. phuEGO processed the user input in two regulation directions
     • increased
     • decreased
  2. In each direction, four levels of phuEGO output:
     • seeds
     • rwr
     • ego (subjected to KDE_cutoff)
     • modules (subjected to KDE_cutoff)
  3. Depending on the levels, three types of files could be provided:
     • protein lists
     • networks (graphml file, use yED to open)
     • fisher's exact test results

• “a user input protein list (seeds) is divided into two directions using the scores (e.g., log2FCs). On each direction, it is expanded through network propagation into a very large list (rwr). Then this list is narrowed down (depending on where the user set the KDE_cutoff) by extracting only Ego network of seed nodes (ego), and further filtered by keeping inter-linked egos (modules -> final result for phuEGO)”

• Main output: Module Network -> pretty much what we need for SPRAS

  • format can be 'edgelist', 'pajek', 'ncol', 'lgl', 'graphml', 'dimacs', 'gml', 'dot', 'leda'
  • files contains redundant information of the module networks in a more human-readable format (csv files)
    • edge_lists and node_attributes
  • Ego network: small networks comprising active functional signalling modules -> also can be in list formats

[agitter]

The most important information is

the feature using own network will be added but not sure when

Until the software supports custom input networks, we can't add it to SPRAS. Let's leave this issue open to monitor the software development. As a new algorithm, it is likely to be updated.