Team:BCCS-Bristol/Modeling

From 2009.igem.org

(Difference between revisions)

Revision as of 18:26, 2 September 2009

BCCS-Bristol
iGEM 2009

Links

Modelling ideas
[http://code.google.com/p/bsim-bccs/ Google Code subversion repository]

Todo list

Steve

Abstract for ECCS (1st Sep)
Analyse performance using TPTP, rerun BSim1.0 simulations w/o various collisions to assess statistical importance
Identify the most interesting parameters over which to plot a quality function
Await details of Paris team GRNs

BSimBacterium: study whether 3D implementation of newPosition in BSimBacteriaCreate is reasonable, tumbleSpeed looks a bit suspect
Reimplement shortcut methods for creating sets of beads/bacteria in BSimParameters
Alllow creation of different types of bacteria again
Growth, vesiculation
BSimBoundingBox: Boundaries in updatePosition()

BSimParameters: add paper references to parameter values in the code using page from last year?
use jYAML for BSimParameters?
Check access control

"Growth curves were measured for all of the mutants, and their log-phase doubling times were calculated. In general, more extreme vesiculation phenotypes corresponded to longer doubling times" (Outer Membrane Vesicle Production by Escherichia coli Is Independent of Membrane Instability)

Emily

~~Issue with bacteria 'escaping' the boundaries~~ r61
- When the bacteria have a large enough force and the time step is not small enough the bacteris are able to 'escape' the boundary.
- Fix this by checking the distance and direction of all the bacteria close to the boundary. This will catch those that have crossed the boundary in one time step.
- THIS COULD BE IMPROVED FURTHER - it allows for far greater forces however does not fix this problem in all cases.

~~Understand how bacteria "tumble" under magnetic force~~
- It is believed that the magnetic force causes the bacteria to orientate in line with the field and so the tumble phase will not occur.

Implement directed bateria (bacteria under constant magnetic field)
- ~~Control only the direction of the bacteria - not the force.~~ r116
- Add some variation in the direction of each bacterium
  - The average alignment of the population is described by Langevin function for classical paramagnetism. [http://arjournals.annualreviews.org/doi/pdf/10.1146/annurev.mi.36.100182.001245]
  - The interaction of the population with a magnetic field is affected by the temperature.

Finish coding the magnetic force for variable magnetic force
- ~~Code the magnetic force as an additional force on the bacteria along with the internal and external forces.~~ r82
- Find realistic values for the magnetic force acting on the bacteria.

Code half-coated bead
- Apply two different potentials to each half of the bead - one side has a potential well (the bacteria attach here) and the other side has no well (the bacteria will interact normally here).

Control which objects are affected by the magnetic force
- Magnetotactic bacteria in a constant magnetic field
- Magnetotactic bacteria in a variable magnetic field
- E. coli attaching to a magnetic bead under a variable magnetic field

Antos

BSim GRNs
- BSimChemicalField generalisation and better integration with scene, rendering etc.
- ~~Proof of concept to check functionality.~~ [http://www.pnas.org/content/101/30/10955 Something like this]
  - Update: it works! (r140) Need to run large scale simulation to check for long term synchronization...
  - Compare chemical field with degradation to vesicular transport.
- ~~Java implementation of Runge-Kutta 4th order solver.~~ r70
  - ~~Refactor ODEs from an interface to an abstract class if necessary.~~ EDIT: interface seems sufficient so far, however need to generalise ODEs to just one interface.
- ~~Investigate other options in terms of external libraries (eg odeToJava - good but seems overcomplicated for current purposes; hundreds of lines of code for one solver routine)~~done
- ~~Implement the solvers into BSim.~~ Done for single bacterium
- ~~Implement other solvers (more efficient).~~ r70
  - ~~The ability to choose between solvers.~~ This will be in a parameter file or some equivalent settings when importing an ODE file.
- Extend to Stochastic ODEs. However, how much stochasticity is inherent in our system? autoinducer chemical field is intrinsically random due to motion of bacteria.
- Interface BSim with external parameters (maybe similar to current parameter files) used to define an ODE system.
  - ~~Investigate the feasibility of SBML parameters or a similar XML based format.~~ SBML may be overcomplicated for our current needs. Low priority for now.
  - Similarly investigate the format used by XPP (may be more succinct, also is specifically for ODEs).
  - If we use JYaml for parameters this may be possible through that.
- ~~Investigate and implement GRN (ODE) and chemical field interaction.~~ r139
  - ~~Study implementation of 3D diffusion in BSim.~~ Seems to work fine
  - ~~Implement diffusion in/out terms for membrane diffusion.~~ r139 - diffusion in/out based on the two relative concentrations
- GRN interaction with vesicle budding and chemical transport (on the surface of the vesicle and inside it).
- Incorporate a method for seeing the effects of GRN activity.
  - ~~Colour changes~~ works but needs reimplementation in new renderer
  - Time series: could have pop-out, need to fix the data export to allow exporting a time series.

New and updated BSim documentation?

BSim graphics
- ~~New rendering framework~~ r182
  - Add boundaries when they're implemented again.
  - Re-implement chemical field(s).
  - Why on earth does the rod bacteria have both radius and diameter parameters?!
- ~~Rod shape rotation.~~ r125
- ~~Basic BSimChemicalField drawing in 3D - will help with grns with a diffusing chemical.~~ r124
  - ~~but needs to be improved in terms of speed~~ Done (removed lighting on chemical field :-s) this is pretty much the limit of optimisations possible without more in depth OpenGL work.
- ~~GRN/quorum field~~ Done, but needs visual improvements due to high dynamic range of chemical levels.
- ~~Investigate (OpenGL?) volume rendering (Tom - Vidi?) maybe better for arbitrary (GRN diffusion) chemical fields~~ A definite possibility in OpenGL, probably possible in P3D for (large?) speedups.

GRNs and vesicles
- Read more about the mechanics of different GRNs (specifically switches).
- Find out how they interact with the external environment.
- ~~Investigate the possiblity of using a different time-step to the fixed one in BSim.~~ Can use a longer or shorter time-step if required, however need to finish other parts to see if this would be relevant/important.
- Investigate the effect of different time steps (GRNs operate on a time scale relatively long compared to that of BSim).
- Investigate the mechanics of our GRNs with respect to vesicle budding and communication.
- Investigate methods for numerically solving stochastic ODEs.

Mattia

Tomski

BSimBatch
- Update to incorporate refactoring of other classes
BSimExport
- What information should be output (numerical data, visualisations, etc)
- Options for visualisation - multiple output cameras, rotation of single camera, following of BSimParticle
- Add options into GUI so that longer simulations can be saved more easily from user interface
- Include new parameters in parameter file and BSimParameters
Compilation on BlueCrystal
- Test jar file generated on BCCS workstation on BC
- Find out how to compile code from command line, without dependancy on Eclipse

@@ Line 60: / Line 60: @@
 === Antos ===
 *BSim GRNs
+** BSimChemicalField generalisation and better integration with scene, rendering etc.
 ** <s>Proof of concept to check functionality.</s> [http://www.pnas.org/content/101/30/10955 Something like this]
-*** Update: it works! Need to run large scale simulation to check for long term synchronization...
+*** Update: it works! (r140) Need to run large scale simulation to check for long term synchronization...
-*** If Garcia-Ojalvo results can be reproduced, compare chemical field to vesicles.
+*** Compare chemical field with degradation to vesicular transport.
 ** <s>Java implementation of Runge-Kutta 4th order solver.</s> r70
 *** <s>Refactor ODEs from an interface to an abstract class if necessary.</s> EDIT: [https://2009.igem.org/Team:BCCS-Bristol/Modeling/Antos/ODE_intefaces interface] seems sufficient so far, however need to generalise ODEs to just one interface.
@@ Line 69: / Line 70: @@
 ** <s>Implement other solvers (more efficient).</s> r70
 *** <s>The ability to choose between solvers.</s> This will be in a parameter file or some equivalent settings when importing an ODE file.
-*** Investigate possibility of using javax/vecmath and matrix ops to maybe make the larger routines more efficient. EDIT: having run ~500 bacteria on RK45 with 7 ODEs each and chemical interaction, it seems that there are other bottlenecks in performance at the moment so this is a very minor issue.
 ** Extend to Stochastic ODEs. However, how much stochasticity is inherent in our system? autoinducer chemical field is intrinsically random due to motion of bacteria.
 ** Interface BSim with external parameters (maybe similar to current parameter files) used to define an ODE system.
 *** <s>Investigate the feasibility of SBML parameters or a similar XML based format.</s> SBML may be overcomplicated for our current needs. Low priority for now.
 *** Similarly investigate the format used by XPP (may be more succinct, also is specifically for ODEs).
-** Investigate and implement GRN (ODE) and chemical field interaction.
+*** If we use JYaml for parameters this may be possible through that.
+** <s>Investigate and implement GRN (ODE) and chemical field interaction.</s> r139
 *** <s>Study implementation of 3D diffusion in BSim.</s> Seems to work fine
-*** Implement diffusion in/out terms for membrane diffusion.
+*** <s>Implement diffusion in/out terms for membrane diffusion.</s> r139 - diffusion in/out based on the two relative concentrations
 ** GRN interaction with vesicle budding and chemical transport (on the surface of the vesicle and inside it).
-** Incorporate a method for seeing the effects of GRN activity (eg colour changes, pop-out time series).
+** Incorporate a method for seeing the effects of GRN activity.
+*** <s>Colour changes</s> works but needs reimplementation in new renderer
+*** Time series: could have pop-out, need to fix the data export to allow exporting a time series.
 *New and updated BSim documentation?
 *BSim graphics
+**<s>New rendering framework</s> r182
+*** Add boundaries when they're implemented again.
+*** Re-implement chemical field(s).
+*** Why on earth does the rod bacteria have both radius and diameter parameters?!
 **<s>Rod shape rotation.</s> r125
-**<s>Basic BSimChemicalField drawing in 3D - will help with grns with a diffusing chemical.</s> Done (r124)
+**<s>Basic BSimChemicalField drawing in 3D - will help with grns with a diffusing chemical.</s> r124
 *** <s>but needs to be improved in terms of speed</s> Done (removed lighting on chemical field :-s) this is pretty much the limit of optimisations possible without more in depth OpenGL work.
 **<s>GRN/quorum field</s> Done, but needs visual improvements due to high dynamic range of chemical levels.
-**BSimChemicalField for 3 chemotaxis fields. Low priority as we're not using more than 1 at a time.
 **<s>Investigate (OpenGL?) volume rendering (Tom - Vidi?) maybe better for arbitrary (GRN diffusion) chemical fields</s> A definite possibility in OpenGL, probably possible in P3D for (large?) speedups.