posterior – Posterior distribution

This module provides a class for representing and working with posterior distribution.

class pebl.posterior.Posterior(nodes, adjacency_matrices, scores, sortedscores=False)

Class for representing posterior distribution.

Except for trivial cases, we can only have an estimated posterior distribution. It is usually constructed as a list of the top N networks found during a search of the space of networks.

The pebl posterior object supports a list-like interface. So, given a posterior object post, one can do the following:

• Access the top-scoring network: post[0]
• Access the top 10 networks as a new posterior object: post[0:10]
• Calculate entropy of distribution: post.entropy
• Iterate through networks: for net in post: print net.score

Note: a posterior object is immutable. That is, you cannot add and remove networks once it is created. See result.Result for a mutable container for networks.

Creates a posterior object.

adjacency_matrices and scores can be lists or numpy arrays. If sorted is True, adjacency_matrices and scores should be sorted in descending order of score.

consensus_network(threshold=0.3)

Return a consensus network with the given threshold.

entropy

The information entropy of the posterior distribution.