new energy nomalization features

[teresa-m]

Test if normalized Energy features perform better to the target. For comparison and selected we want a feature ranking list and the feature correlation plot.

TODO:

• [ ] compute new energy features
  • [ ] E/ sum(target, query interaction lenght)
  • [ ] EDs/ sum(target, query interaction lenght)
  • [ ] E-hybrid/ sum(target, query interaction lenght)
  • [ ] E/ sum(target, query interaction lenght) + (GC-content*interaction length)
  • [ ] EDs/ sum(target, query interaction lenght) + (GC-content*interaction length)
  • [ ] E-hybrid/ sum(target, query interaction lenght) + (GC-content*interaction length)
    • [ ] calculate feature ranking
    • [ ] check if spearman correlation is from -1 to 1
    • [ ] plot feature correlation

[teresa-m]

Not sure how to best combine the length and GC content normalization