Package transsys :: Module optim
[hide private]
[frames] | no frames]

Module optim

source code

Classes [hide private]
  Interval
Represents an interval [a, b].
  FitnessResult
Base class for results of objective (or fitness) functions.
  LsysDisparityFitnessResult
Result of a disparity objective evaluation.
  Mutator
  TransformationFunction
Base class for transformation functions.
  ExponentialFunction
Exponential parameter transformation.
  ArctanFunction
Arcus tangens transformation function.
  LowerBoundedFunction
A function asymptoting to a finite lower bound as x -> -Inf and linearly diverging as x -> +Inf.
  SigmoidFunction
Sigmoid parameter transformation.
  ParameterTransformer
Base class for parameter transformers.
  TranssysTypedParameterTransformer
Parameter transformer differentiating decay, diffusibility, constitutive, aspec, amax, rspec and rmax nodes.
  OptimisationRecord
Base class for optimisation records.
  SimulatedAnnealingRecord
Record containing data from a Simulated Annealing update.
  GradientOptimisationRecord
Record containing data from a step of gradient optimisation.
  OptimisationResult
Base class for returning results of an optimisation.
  SimulatedAnnealingResult
Result of optimisation by Simulated Annealing.
  GradientOptimisationResult
Result of optimisation by Simulated Annealing.
  AbstractObjectiveFunction
Base class for objective functions.
  ExpressionSeriesObjective
Abstract base class for objective functions based on comparing observed with desired expression profiles.
  ExpressionSeriesSquareSumObjective
The sum of squared Euclidean distances to a desired set of expression profiles as an objective function.
  ExpressionSeriesCorrelationObjective
Correlation based objective function.
  AbstractOptimiser
Abstract base class for optimisers.
  SimulatedAnnealer
Simple Simulated Annealing tool.
  GradientOptimiser
Gradient descent optimiser.
  LsysObjectiveFunction
Functions [hide private]
 
active_and_inactive_set(rule, instance_series, factor_index)
Extract the set concentrations of a factor in instances that activated a rule (the active set), and the complement set (the inactive set).
source code
 
stddev_disparity(rule, instance_series, factor_index)
Standard deviation based disparity score.
source code
 
difference_sign_disparity(rule, instance_series, factor_index)
Disparity of sign values of differences between expression levels in symbols activating rule and other symbols.
source code
 
overlap_disparity(rule, instance_series, factor_index)
Disparity of expression of factor in symbols activating rule and symbols not activating rule.
source code
 
flat_symbol_instance_list(lstring_list, transsys_program=None) source code
 
disparity_fitness(lsys_program, transsys_program, factor_names, num_timesteps, disparity)
Objective function quantifying disparity of expression levels between instances that activate and that do not activate rules.
source code
 
hillclimb(objective_function, decoder, initial_dnaseq, num_generations, mutator)
Experimental function for sequence encoding.
source code
 
get_value_nodes(transsys_program, gene_name_list, factor_name_list)
Get value expression nodes from a transsys program.
source code
 
randomise_transsys_values(transsys_program, random_function, gene_name_list=None, factor_name_list=None)
Randomise values in value expressions.
source code
 
parse_transformation_function(f) source code
ParameterTransformer
parse_parameter_transformer(f)
Read a parameter transformer from a file.
source code
 
parse_optimiser(f) source code
Function Details [hide private]

active_and_inactive_set(rule, instance_series, factor_index)

source code 

Extract the set concentrations of a factor in instances that activated a rule (the active set), and the complement set (the inactive set).

This method is rather specific to L-systems optimisation and should perhaps not be in this module.

stddev_disparity(rule, instance_series, factor_index)

source code 

Standard deviation based disparity score.

This function applies some tricks for preventing limitations of floating point precision from resulting in artificially low scores. See comments in source for details.

difference_sign_disparity(rule, instance_series, factor_index)

source code 

Disparity of sign values of differences between expression levels in symbols activating rule and other symbols.

Evaluates to 0 if all signs are the same (and hence activating and inactivating instances have expression levels of the factor specified by factor_index in non-overlapping intervals and to 1 if sign ratio is 50:50.

disparity_fitness(lsys_program, transsys_program, factor_names, num_timesteps, disparity)

source code 

Objective function quantifying disparity of expression levels between instances that activate and that do not activate rules.

This should be rewritten as a proper AbstractObjectiveFunction subclass.

get_value_nodes(transsys_program, gene_name_list, factor_name_list)

source code 

Get value expression nodes from a transsys program.

Deprecated: use methods of TranssysProgram.

randomise_transsys_values(transsys_program, random_function, gene_name_list=None, factor_name_list=None)

source code 

Randomise values in value expressions.

Deprecated: This functionality is now integrated into the optimising framework.

parse_parameter_transformer(f)

source code 

Read a parameter transformer from a file.

This function uses the "magic" first line to determine what subtype of ParameterTransformer is specified.

Returns: ParameterTransformer
the parameter transformer