Assume a network as a set of data, which are defined by their coordination $(x,y,z)$ and a weight on its edge. Now this data can be used as an input data to predict a single value.

In my case, coordination of nodes are mostly fixed, and just by changing subset of node I would like to predict the output.

For example, in protein structure, coordination of each amino acid is given (nodes) and atomic bond can be calculated based on physical property and distance of each amino acid,and we would like to see by changing couple of amino acids (either by displacement or replacing by some other amino acid) predict the pH or solubility of the protein which are single values.

Also, in gene-gene interaction network, one can do more or less same thing by changing some genes, predict the drug performance and side effect.

I appreciate any keyword or literature for such a method.

  • $\begingroup$ Perhaps you could clarify this question by indicating how the data are used to predict a value (and what value is being predicted). What do you mean by "changing subset of node"? What is the "output" of a network? $\endgroup$ – whuber Sep 12 '11 at 22:07
  • $\begingroup$ how is that now ?! $\endgroup$ – user4581 Sep 13 '11 at 8:38
  • $\begingroup$ This sounds like a question of computational chemistry rather than statistics or machine learning. Abstracting the situation down to a generic "network" loses almost all the important information. The prediction is a matter of scientific modeling. $\endgroup$ – whuber Sep 13 '11 at 13:12
  • $\begingroup$ Try "learning Bayesian networks". $\endgroup$ – JohnRos Dec 22 '11 at 7:40

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.