Skip to content

Latest commit

 

History

History
166 lines (106 loc) · 6.58 KB

README.md

File metadata and controls

166 lines (106 loc) · 6.58 KB
Raw

MASTERS

HOW TO OPEN IT ON NETLOGO?

  1. First of all you need NETLOGO. Make sure you have the Netlogo matrix, array and shell extensions installed. You have 2 options:
    1. Download Netlogo from it website: http://ccl.northwestern.edu/netlogo/ and then download the extensions and install them. You will need to download Masters_vBETA.nlogo3d too.
    2. Download our netlogo folder, with MASTERS inside (along with all other netlogo models) and with the extensions already installed. You will find it here in github, at the folder Netlogo-MASTERS.
  2. Execute the netlogo-3d.sh script. (Linux) or Open the NetLogo 3D 5.0.4 executable (Mac).
  3. Click FILE -> OPEN then find /MASTERS/Masters_vBETA.nlogo3d
  4. For fast simulations it is advisable to unmark the "view uptdates" option at the 3D View window.
  5. It's done! Enjoy MASTERS!

WHAT IS IT?

MASTERS is a general sequence-based MultiAgent System for protein TERtiary Structure prediction.

HOW IT WORKS

MASTERS is based on an ab initio approach. It's a cooperative hierarchical multiagent system guided by a combined Simulated Annealing/Monte Carlo scheme to address the PSP problem. The main idea behind MASTERS is to provide the user with freedom to choose the abstraction level, as well as the energy function/force field to lead the simulation.

THINGS TO NOTICE

When running the program the user can look at the Acceptance Ratio plot and see the progress of the simulation. It is recommendable a not so high acceptance ratio, for example.

It is important to stress that (1) type of movements, (2) model abstraction level and (3) energy function are three sides of the same triangle. Changing one of them will affect the behavior of the entire.

THINGS TO TRY

Create or apply the abstraction level of your choice! Create or apply the energy function of your choice! Create or apply the scheme of movements of your choice!

EXTENDING THE MODEL

The code is open, you can dig in and modify the optimization (Simulated Annealing/Monte Carlo) procedure as your wish.

NETLOGO FEATURES

One of the remarkable advantages of Netlogo is its embedded tools,being BehaviorSpace is one of them. BehaviorSpace offers the possibility of automatically performing a big set of experiments based on changing parameters' values. In MASTERS, due to the BehaviorSpace capability, it is possible to explore more efficiently the conformational space of PSP predictions and tune MASTERS's parameters to achieve better results.

Try it yourself!

Go to Tools -> BehaviorSpace. There we have 2 samples:

  1. How to take 2 sets of parameters: Each one run 40 times recording, at the end of each simulation, a lot of attributes including the final x,y and z coordinates and the energy of every tick(timestep).

  2. How to tune all your parameters: In the sample there are 7 different parameters. Each one with at least 2 options/values. Running this experiment will result in a combinatorial number of runs, regarding the combination of all the parameters/values.

HOW TO USE IT

##QUICK GUIDE If you want to quickly test MASTERS, you can use the following step by step:

  1. First of all you need to fill the Fasta-Seq field with a sequence of your interest. Remember that the Netlogo's world size must support the size of your target. (You can change the world size right-clicking the 3-D view)

  2. Click the Startup button (or press Q)

  3. Click the Load Fasta Seq into View (or press A)

  4. Click the Create director and environment agents button (or press S)

  5. Click the Environment Agent - Go button (or press D). For fast simulations it is advisable to unmark the "view uptdates" option at the 3D View window.

SETUP COMMANDS

###title Normally it is the ID of the protein. It will be at the top of the .pdb file, if the user exports the 3-D visualization using the create_pdb_from simulation procedure.

Example:

create_pdb_from simulation "file.pdb"

###nr-of-directors Input--> Number of Director Agents. Default value: 1.

###limit_prob_dir and limit_prob_search Input--> Thresholds related to the acceptance ratio of the Director and Searching agents.

###parameters Input--> Used by the procedure load_parameters to prepare the simulation. Follows the pattern a-b-c-d-e-f-g, where:

a = orchestra-sleep-time;

b = temp_factor_dir;

c = temp_factor_search;

d = nr-of-directors;

e = limit_prob_dir;

f = limit_prob_search

g = attempted_threshold_with_dir

###Resize Button used to change the searching agent's size. It will not affect the simulation, just the 3-D visualization.

###Clear All Clears all the input textboxes.

MOVEMENTS

SEARCHING AGENTS MOVEMENT

SCHEME 1

The scheme 1 movements are based on angles. The distance from two consecutive residues remain the same. This is a requirement of the default energy function (AB Model). There are two types:

  • 1 DOF move(Degrees of Freedom): Most of the residues move in 1DOF, keeping fixed the distance from the residues from the front and back.
  • 3 DOF's: Just the first and last residues move that way. They also keep fixed their distances but they are binded at just one residue, so they have more freedom to move.

SCHEME 2

The scheme 2 movements are based on distance. The agent is placed on the center of a 1A cube and have the same probability to move to each other place on this cube. All the phase space is accessible this way. It is simple to achieve erodicity.

DIRECTOR AGENTS MOVEMENT

CRANKSHAFT MOVE

First choose 2 residues A and B chain (at a distance of 3 residues). Rotate the residues that separate A and B by choosing a random angle, using the line from A to B as axis.

PIVOT MOVE

Choose a point (pivot residue) and a part of the chain rotates around this point using a random angle.

CONTROL MONITORS

Nr of Searching Agents.

Shows the current number of Searching agents

Nr of Director Agents

Shows the current number of Director agents.

Nr of Environment Agents

Shows the current number of Environment agents.

Step

Shows the timestep or temperature step (related to the simulated annealing scheme).

Current Energy

Shows system current energy based on the compute_global_energy procedure.

Lower Energy and Lower Energy tick

Shows the best(lower) obtained energy.

WORKFLOW

To be filled.

PLOTS

To be filled.

ATTEMPTED MOVES CONTROL

To be filled.

SEARCHING AGENTS

To be filled.

DIRECTOR AGENTS

To be filled.

ENVIRONMENT AGENTS

To be filled.

MONTE CARLO FACTORS

To be filled.

DEBUGING

To be filled.

CREDITS AND REFERENCES

PUBLICATION

MASTERS: A general sequence-based MultiAgent System for protein TERtiary Structure prediction

CS2BIO 2014