psamm.fastcore – Fastcore (approximate consistent subset)

Fastcore module implementing the fastcore algorithm

This is an implementation of the algorithms described in [Vlassis14]. Use the functions fastcore() and fastcc() to easily apply these algorithms to a MetabolicModel.

exception psamm.fastcore.FastcoreError

Indicates an error while running Fastcore

class psamm.fastcore.FastcoreProblem(*args, **kwargs)

Represents a FastCore extension of a flux balance problem.

Accepts the same arguments as FluxBalanceProblem, and an additional epsilon keyword argument.

Parameters:
  • model – MetabolicModel to solve.
  • solver – LP solver instance to use.
  • epsilon – Flux threshold value.
lp7(reaction_subset)

Approximately maximize the number of reaction with flux.

This is similar to FBA but approximately maximizing the number of reactions in subset with flux > epsilon, instead of just maximizing the flux of one particular reaction. LP7 prefers “flux splitting” over “flux concentrating”.

lp10(subset_k, subset_p, weights={})

Force reactions in K above epsilon while minimizing support of P.

This program forces reactions in subset K to attain flux > epsilon while minimizing the sum of absolute flux values for reactions in subset P (L1-regularization).

find_sparse_mode(core, additional, scaling, weights={})

Find a sparse mode containing reactions of the core subset.

Return an iterator of the support of a sparse mode that contains as many reactions from core as possible, and as few reactions from additional as possible (approximately). A dictionary of weights can be supplied which gives further penalties for including specific additional reactions.

flip(reactions)

Flip the specified reactions.

is_flipped(reaction)

Return true if reaction is flipped.

psamm.fastcore.fastcc(model, epsilon, solver)

Check consistency of model reactions.

Yield all reactions in the model that are not part of the consistent subset.

Parameters:
  • model – MetabolicModel to solve.
  • epsilon – Flux threshold value.
  • solver – LP solver instance to use.
psamm.fastcore.fastcc_is_consistent(model, epsilon, solver)

Quickly check whether model is consistent

Return true if the model is consistent. If it is only necessary to know whether a model is consistent, this function is fast as it will return the result as soon as it finds a single inconsistent reaction.

Parameters:
  • model – MetabolicModel to solve.
  • epsilon – Flux threshold value.
  • solver – LP solver instance to use.
psamm.fastcore.fastcc_consistent_subset(model, epsilon, solver)

Return consistent subset of model.

The largest consistent subset is returned as a set of reaction names.

Parameters:
  • model – MetabolicModel to solve.
  • epsilon – Flux threshold value.
  • solver – LP solver instance to use.
Returns:

Set of reaction IDs in the consistent reaction subset.
psamm.fastcore.fastcore(model, core, epsilon, solver, scaling=100000.0, weights={})

Find a flux consistent subnetwork containing the core subset.

The result will contain the core subset and as few of the additional reactions as possible.

Parameters:
  • model – MetabolicModel to solve.
  • core – Set of core reaction IDs.
  • epsilon – Flux threshold value.
  • solver – LP solver instance to use.
  • scaling – Scaling value to apply (see [Vlassis14] for more information on this parameter).
  • weights – Dictionary with reaction IDs as keys and values as weights. Weights specify the cost of adding a reaction to the consistent subnetwork. Default value is 1.
Returns:

Set of reaction IDs in the consistent reaction subset.