This mode search seeds by ensuring logical constraints and can be used with accumulation constraints of a metabolite.
📝 Notes:
There is no linear calcul, so the stoichiometry of metabolites are not involved into the algorithm of seed searching, such as the boundaries of a reaction.
After analysing and converting the network a seed resoltuion is performed. Please read the 📜network analyse document to understand the technical terms
The Full Network mode set all metabolite as Target. The targets must be in the scope of producible metabolites by the network.
A metabolite is an external seed when:
- The metabolite is a reactant of a reaction and not a product of any reaction (not exchanged reaction and not reversible reaction). The metabolite is not a transported⭐ metabolite
- The metabolite is not a product of any reaction except his own exchange reaction
- Specific case of reversible transport reaction involving a transported⭐ metabolite without any import reaction. The reversibility prevents Seed2LP to select one of the metabolite as seed because it becomes a product of a reaction. This case is taken into account and the classic metabolite is chossen as seed
💬 Comments:
⭐ A transported metabolite is a metabolite that is transferred from one compartment to another. The formula is therefore generally something like:
$\quad$ $\quad$ $M_A_e ⟷ M_A_c ⟷ M_A_mBecause it the same metabolites on different compartment, the program tags M_A_c and M_A_m as transported metabolite, therefore only M_A_e is selected as seed.
This restriction allows to avoid similar set of seeds such as {M_A_e, M_B} or {M_A_c, M_B} or {M_A_m, M_B}
📃 Example
Using the reaction below, M_A_c is an R_A is an exchange reaction and the the metaoblite M_A_c is set as exchange metabolite and will be chosen as seed because it complies with the seed external definition. M_B_m and M_B_c are both product of the reaction (or the reversible). One of them will be taged as classic metabolite and the other as a transported reaction, then while searching seeds, the classical M_B metabolite will be chosen as seeds
flowchart TD None-->C{R_A}; C-->M_A_c; M_A_c & M_B_c-->R; R{R_1}-->M_D_c; M_B_c-->B{R_B} M_B_m-->F{Rev_R_B} B-->M_B_m; F-->M_B_c;
There is a set of impossible seeds, which are metabolite that are products but never reactant.
SBML source file are sometimes are written with reaction without using the product of this reaction in another reaction, even export reactions. This is an accumulation already existing in the source file.. This type of network can give an unsatisfiable problem. In order to prevent this, Seed2LP forbids by default the accumulation.
The accumulation is forbidden with a constraint:
- a metabolite product wich is in the scope (producible metabolite) must be consummed at least once by a reaction
Because most of the time the network doesn't result to an unsatisfiable problem, it is possible to allow the accumulation by using option --accumulation / -accu. But it is possible that the solution does not ensure flux while testing it with --check-flux / -cf option in a second step.
This prohibition of accumulation is not needed when using filter or guess-check mode because the flux calculation with cobrapy is tested for each solution while searching them.
For the classic mode, no other constraints are added to the Algorithm. All solutions are selected and outputted without any flux verification.
No other constraints are added to the Algorithm, but for all solutions the flux is checked with cobrapy, and only the solutions having flux are selected and outputted.
This mode directly interact with the solver during solving by adding new constraints to the Algorithm. Flux is checked with cobrapy for every solution proposed by the solver and new constraints are derived out of it to guide the remaining solving process. All outputted solutions have flux natively.
This mode search seeds by ensuring logical and linear constraints and the accumulation option is not taken into account if used.
the Logic part of the algorithm is the same as the reasoning algorithm, and add some logic constraints in order to complete linear calcul.