Neighbour's list

Time to add a neighbour's list to the Measure concept.
A trait class (proposed name: Neighbours) should be written for this purpose.

The collision solver should examine the neighbours list for the persistence of collision pairs.

The current collision solver is too glue-some that once a participant is added, it cannot go unless the whole thing dissociates.

To do on Monday (again)

1) Complete Trajectory::WriteInfo() routine with config information.
2) Complete Trajectory::Read() member for reading traj dirs.
3) Modify biosimzstats, biosimzapp and biosimzlab to adapt changes.
4) Test FileAttached class for path manipulations in BioSimzTest.
5) Test Trajectory in BioSimzTest.
6) Test all file-related functions are working normally.
7) Test CollisionSheet in simulations.

To do (Tuesday on):

8) Modify submission scripts adapting the changes made to trajectory management.
9) Set up simulations for barnase/barstar and ras/effectors.
10) Write code for parsing CollisionSheet into tables/svg graphs.
11) GUI improvement and manuscript preparation.

Also pending:

12) Prepare a figure for Paul's annual report on BioSimz.

!!!!!!!!!!!!!!

To do on Monday

Add event counter system to the collision solver to be able to serialise the binding event history of a trajectory. This is for future expansion on biosimzstats regarding reaction rate calculation.

Further investigation on how to calculate the rate constants (wrt BioSimz) needs to be carried out.

It has been suggested that for known interactions in MolView data plots, the centre-of-mass of the interface residues should be plotted instead of that of the whole protein.

Literature search for the "sticky" nature of protein-protein interactions should be performed.

To do in the evening

To do:

a) modify the variable name m_effDist to m_maxPtl in vdw and hbond forcefields
b) increase the vdw attraction to stabilise dynamic interactions.

Parameters for the OLD vdw settings

min distance: 2.99
attractive scale: 2.3

To do: the desolvation forcefield

Claimed in Camacho's paper it is the major factor, besides electrostatic forces, that governs or helps protein binding, elevating the k-on constant by several orders of magnitude.

Inspired on their implementation I should realise the desolvation forcefield in this way

a) find the surface patches that are less than 4.2A in distance;
b) add the area of the surface to a total value;
c) proportionate this value to energy;
d) find the unit vector of this virtual desolvation force;
e) calculate the force.

Before doing this, a robust reading-from/writing-to-file routine should be written to automate the surface meshing to save computing time/power.

To do list

1. Mark the low-RMSD points
(a) Get the correct Complex object
(b) Push the target pair into a container
(c) Use DockChecker to get RMSD
(d) If it falls within threshold, then mark it with a special colour
2. Research into the BSP tree and other collision facet-detection algorithms
3. Designing the interface to fire up more runs

Revision 1000

I set up the BioSimz project on SourceForge in March 2007.
Till just now the repository has accumulated 1000 revisions.

Cont'd 2

The real problem turned to be a bug in the MemPool of which the initialisation with 0 size shall not be allowed. The bug has been rectified and thus the previous statements are no longer valid.

However, there are serious stability issues (seg faults/mem violations) whenever something is pushed into the m_objVis container in the WnMolView class, regardless of the container's type.

Cont'd

Maybe it's not the general container's problem. It looks as if the GLTextList class cannot survive a copy construction -- most probably because of the copy construction of LinkedList<>.

Default copy constructors

Do not rely on compiler-generated copy constructors for objects which have container members such as Container<>, LinkedList<>. The copy constructors of those containers have unsafe copy routines, at the benefit of type-tolerance, that could result in memory violations.

Further improvements can be made to introduce an internal push_back() operation that is strictly typed (non-templated) to eliminate this imcompatibility.

Versions

Last stable version without molecular paths: rev 985.
New code for molecular paths (unfinished): rev 986/987.

The (finally coming) MolView in BioSimzLab

Feature/task list

• Have the MolInfo window fully linked to MolView, including triggering launch of the MolView window, as well as identifying existing ones.
• The MolView should have central QGLViewer widget like the SimReplay one without the tool bar for play control. Instead it should have 3 small QGLViewer widgets aligned to the top right of the window providing views for x-y, y-z, and x-z planes
• A list with selection boxes/or "*" marks should be placed beneath the 3 small GL widgets. It is a list of all the data layers plotted in the view, including the molecular surface, any traces/paths, etc. Double-clicking each items should toggle the display of that layer on/off.
  

To do

1. Add the display list for highlight mode in surface mode
2. Link the GL selection to the mdl selection
3. Create the molecule analysis window

BioSimzLab - display re-organisation

Some thoughts

• Have the simulation replay window based on QGLViewer
• Molecules clickable and linked to new tabs opened for individual analysis
• Redirect the stdout to a textbox
  

libQGLViewer to be assesed

Have found this library possibly useful to upgrade the current GLViewer in BioSimzLab.

Feature planned: docked simzprofile windows

Description: Docked windows listing the profile information
Class: WnTplList, WnAgentList, WnSimzInfo, WnSimzRun
Dependency: BioSimz, QtGui, QtXml

Feature planned: heat map projection

Description: Heat maps projected onto protein surfaces.
Class: Heatmap
Dependency: BIOSIMZ::Color, CGAL::Polyhedron_3

Set-up!

In this blog I will write in as much detail as I can (or, will) about the BioSimz project, a C++ library for conducting large-scale biomolecular simulations.