class modifinder.engines.evaluation.BasicEvaluationEngine(default_method='average_distance', **kwargs)

Class to evaluate the probabilities based on the basic evaluation engines explained in the paper

Parameters:

• default_method (str or None (default = "average_distance")) –

  The default evaluation method to use. If None, the user must specify the method. List of methods:

  • is_max : returns 1 if the true modification site is the atom with the highest probability, 0 otherwise

  • proximity : returns the proximity to the atom with the highest probability

  • average_distance : returns the average distance to the atom with the highest probability normalized by the graph diameter

  • sorted_rank : returns the rank of the true modification site in the sorted probabilities

• kwargs (dict) – Additional arguments.

evaluate(network, **kwargs)

Evaluates the network.

Warning

Not implemented yet.

Parameters:

• network (type) – Description of the network parameter.

• **kwargs – Additional keyword arguments.

evaluate_single(known_compound_structure, unknonwn_compound_structure, probabilities, evaluation_method=None, **kwargs) → float

Evaluates the probabilities based on the evaluation method

Parameters:

• known_compound_structure (rdkit Chem.Mol) – The known compound structure

• unknonwn_compound_structure (rdkit Chem.Mol) – The unknown compound structure

• probabilities (List[float]) – The probabilities of each atom being the modification site

• evaluation_method (str) – The evaluation method to use. If None, the default method will be used.

• kwargs (dict) – additional arguments

Returns:

The evaluation score based on the evaluation method. The score is between 0 and 1.

Return type:

float