Giters

hooman650 / BioModelChecker

Bio-ModelChecker: Using Bounded Constraint Satisfaction to Seamlessly Integrate Observed Behavior with Prior Knowledge of Biological Networks

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

General Requests and Enhancements

hooman650 opened this issue · comments

commented

Feature Request and enhancement

This is an attempt to track the enhancements and feature requests in BioModelChecker 🔉 🔬
=========================

  • If a model only contains Steady State constrains and no Trajectory, BioMC does not import its solutions properly (reported by JohnC)
  • Import solutions from multiple folders
  • Check the input json file and check whether length of Trajectory matches the number of entities
  • Allow constraints for "rest" value of each node (decay)
  • Match states in output trajectories to input (which states are added automatically by model checker?)
  • Select SS1 or SS2 from competing models as MIS targets
  • The tool does not run properly with its shortcut desktop version (e.g. path-issue)
  • Parse K matrix to display truth tables for each node (e.g. Output a multivalued logical equation)
  • Identify available node-steady states supported by a given parameter set (SS detection)
  • Identify oscillatory attractors
  • Identify failed constraints (specific causes of unsatisfiability)

=========================
🚣 🚛

commented

@mcmrgh Find the latest release, in the release tab.

commented

It looks like solutions cannot be imported if the initial json didn't have a Tr field. This can be remedied by adding a "dummy" trajectory field to the json, but it's a little clumsy for processing solutions.

commented

Possible to estimate total search space size and keep track of progress? Complex models can take a long time to converge, and it would be useful to get a rough idea of how much search space remains at regular intervals.

commented

It looks like solutions cannot be imported if the initial json didn't have a Tr field. This can be remedied by adding a "dummy" trajectory field to the json, but it's a little clumsy for processing solutions.

@mcmrgh Thanks for letting me know. I will resolve this soon. Meanwhile, you will be able to import the solutions by selecting the "Pathlength" as the option from the main panel prior to the import.

commented

Possible to estimate total search space size and keep track of progress? Complex models can take a long time to converge, and it would be useful to get a rough idea of how much search space remains at regular intervals.

@mcmrgh Possible to estimate the search space but constraint solvers search the constraint space. I am afraid this is "solver" dependent, some solvers provide indications in regards to the process (e.g. traversal of the constraint graph) some don't. Besides as the problem is NP-hard and also multi-objective, there is no guarantee that an optimal solution would be found at all (If any).

commented

Solution db does not include steady states, though they can be extracted from text output.

commented

I ran MIS discovery on a set of 40 models for 3 hours each; it progressed well (with MIS identification and robustness checking) through the first 4 models but then hung. Log shows only that the Matlab parallel pool was shut down after an idle period.

commented

Is it possible to constrain trajectories that do not exist?
i.e. if a node is 0, there is no way for it to become 1 without external intervention, though decay from 1 to 0 is valid.

commented

Some metric for complexity of K values would be very helpful for assessing the quality of solutions-- when solving exhaustively, many "trivial" solutions are often generated by alteration of a K value which is seldom invoked.