Exercise 2: Meta‑Networks

Prof. Patrick E. Meyer

Using the following file, containing 5 networks: networks.txt located in the public/biol0021/ directory (also available from the course webpage):

  1. Load the different networks (using the function read.table()).
  2. Replace the missing edges in all the networks by a standard “I don’t know” value of 0.05.
  3. Find the inferred network that is the most correlated to redfly.
  4. Is a meta‑network weight‑sum improving the results?
  5. Based on the top 5 % edges of the best predictive network, can you suggest a few genes that could be targeted in order to alter the pathway in which FBgn0001180 is involved?
    1. First, select the top 5 % edges of the best network (the most correlated to redfly) and make it an igraph object.
    2. Plot the induced subgraph made of the target gene and all its neighbours.
    3. Compute the z‑score of the eigen‑vector and betweenness centrality measures for our target gene.
    4. If one of those scores is too high, then compute those measures for all its regulated neighbours and suggest some “not‑too‑important” genes to target.