Blender in Bio-/ Quantum-chemistry. Thomas Haschka - Blenderconf 2011

Transcription

1 Blender in Bio-/ Quantum-chemistry Thomas Haschka - Blenderconf 2011 thomas.haschka@inserm.fr

2 Our Mission Academic research institution Mixed biology/chemistry/physics/ informatics interdisciplinary environment Molecular dynamics / quantum chemistry etc. Building/using mathematical models to explain biological processes on a molecular level

3 What does it take to explain a biological process? A biological process happens at different scales. Different scales require different kinds of treatments and careful linkage between them.

4 What are our Scales?

5 Proteins Continuum >10nm Building Blocks: Entire Biological Molecule (Protein) Atoms Blocks of such a Biological Molecule Methods: Mechanical Models Fluid Dynamics Langevin/Brownian Dynamics Electrons

6 All Atoms ~0.1nm Proteins Coarse Grained Building Blocks: Atoms Atom Assemblies Methods: Molecular Dynamics Simulations CG MD Simulations Langevin/Brownian Dynamics Electrons

7 What is classical molecular dynamics? Atoms have modeled interactions which are trying to describe reality. Correctness vs (amount of data and computational effort) Springs electrostatics

8 Proteins 0.1 nm > Quantum Building Blocks: Subatomic Electrons - Wave function Methods: Quantum Mechanical Mixed Quantum/classical molecular dynamics simulations Electronic Shell Evaluations Electrons

9 Why Blender? Allows us to visualize our molecules in three dimensions. At all scales the molecule s function is determined by its three dimensional structure. Visualizing a molecule thus means in many cases that you know how it actually works! Visualization further helps you to communicate your discoveries.

10 What else besides Blender? Different scales require different programs What we are using: Hand written code, various tools,... GROMACS - Molecular Dynamics GAUSSIAN - Quantum Mechanics VMD, PyMOL generate 3D VRML models Blender to show off our results!

11 Workflows That s how we do it!

12 Continuum Mechanics

13 Continuum Mechanics Create a mathematical model of your molecule Create a visual model of your mathematical model

14 Continuum Mechanics Create a mathematical model of your molecule Create a visual model of your mathematical model

15 Continuum Mechanics Create a mathematical model of your molecule Simulate Create a visual model of your mathematical model

16 Continuum Mechanics Create a mathematical model of your molecule Simulate Create a visual model of your mathematical model

17 Continuum Mechanics Create a mathematical model of your molecule Simulate Create a visual model of your mathematical model Animate your visual model

18 Continuum Mechanics Create a mathematical model of your molecule Simulate Create a visual model of your mathematical model Animate your visual model

19 Continuum Mechanics Create a mathematical model of your molecule Simulate Create a visual model of your mathematical model Animate your visual model Render

20 MD Simulations

21 Obtain / model a structure of a Protein. Optionally generate a coarse grained model MD Simulations

22 Obtain / model a structure of a Protein. Optionally generate a coarse grained model MD Simulations

23 MD Simulations Obtain / model a structure of a Protein. Optionally generate a coarse grained model Simulate Create a visual model of your structure.

24 MD Simulations Obtain / model a structure of a Protein. Optionally generate a coarse grained model Simulate Create a visual model of your structure.

25 MD Simulations Obtain / model a structure of a Protein. Optionally generate a coarse grained model Simulate Analyze your simulations and interpret them. Create a visual model of your structure.

26 MD Simulations Obtain / model a structure of a Protein. Optionally generate a coarse grained model Simulate Analyze your simulations and interpret them. Create a visual model of your structure.

27 MD Simulations Obtain / model a structure of a Protein. Optionally generate a coarse grained model Create a visual model of your structure. Simulate Model features you find to be important into the model of your structure. Analyze your simulations and interpret them.

28 MD Simulations Obtain / model a structure of a Protein. Optionally generate a coarse grained model Create a visual model of your structure. Simulate Model features you find to be important into the model of your structure. Analyze your simulations and interpret them.

29 MD Simulations Obtain / model a structure of a Protein. Optionally generate a coarse grained model Create a visual model of your structure. Simulate Model features you find to be important into the model of your structure. Analyze your simulations and interpret them. Render

30 Quantum Chemistry Obtain / model a structure of your molecule Calculate quantum features of your molecule (like the electron density) Generate volumetric data from your obtained results Render

31 Other Scientific Tasks Arbitrary Volume Data Function Plotting Molecular Properties Structural Alignment Normal Mode Analyses Phase Space Visualization

32 Case Study: Thrombospondin C-Terminal

33 TSP Model in Blender C-Terminal/ Signature TSR1-Repeats CC-Region Quickly modeled (like 20 minutes) in Blender using just spheres and Bezier curves. Allows us to point out different regions of the molecule, and those we are interested in. N-Terminal Works way better then hand drawn 2d schemes.

34 Signature Domain (Atomistic Level) Mesh has been generated in Pymol from a structure available in the Protein Data Bank Around 40, 50ns all atom simulations have been evaluated. We used blender to highlight main features of the structure known from literature. Further we used armatures to model major motions found by the MD simulations.

35 Stalk Move Calcium Ion Integrin Binding Site CD-47 Globe Binding Site Still Frame with Important Sites

36 Structural Alignment Different members of the TSP family share the same signature domain TSP-1 TSP-2 TSP-5

37 Quantum Mechanics Many movements, biological processes can not be described by the simplicity of classical molecular dynamics. In this cases we need to use data evaluated by the principles of quantum mechanics. In the case of thrombospondin the electron cloud around calcium binding sites was evaluated using GAUSSIAN. GAUSSIAN allows us to express its results in so called cube files, which essentially is voxel data.

38 So far no classical molecular visualization software (VMD, PyMOL) allows us to visualize this data at an arbitrary precision.

39 Getting Quantum Data into Blender

40 Getting Quantum Input Data into Blender

41 Getting Quantum Input Data into Blender

42 Input Gaussian/ cubegen Getting Quantum Data into Blender

43 Input Gaussian/ cubegen Getting Quantum Data into Blender

44 Getting Input Gaussian/ cubegen cubes Quantum Data into Blender

45 Getting Input Gaussian/ cubegen cubes Quantum Data into Blender

46 Getting Input Gaussian/ cubegen cubes Quantum Data into Blender cubealign

47 Getting Input Gaussian/ cubegen cubes Quantum Data into Blender cubealign

48 Getting Input Gaussian/ cubegen cubes Quantum Data into Blender cubealign coords

49 Getting Input Gaussian/ cubegen cubes Quantum Data into Blender cubealign coords

50 Getting Input Gaussian/ cubegen cubes Quantum Data into Blender cubealign cubegen coords

51 Getting Input Gaussian/ cubegen cubes Quantum Data into Blender cubealign cubegen coords

52 Getting Input Gaussian/ cubegen cubes Quantum Data into Blender cubes cubealign cubegen coords

53 Getting Input Gaussian/ cubegen cubes Quantum Data into Blender cubes cubealign cubegen coords

54 Getting Input Gaussian/ cubegen cubes Quantum Data into Blender cubes cubealign cube2raw cubegen coords

55 Getting Input Gaussian/ cubegen cubes Quantum Data into Blender cubes cubealign cube2raw cubegen coords

56 Getting Input Gaussian/ cubegen cubes Quantum Data into Blender cubes cubealign cube2raw cubegen coords raw

57 Getting Input Gaussian/ cubegen cubes Quantum Data into Blender cubes cubealign cube2raw cubegen coords raw

58 Getting Input Gaussian/ cubegen cubes Quantum Data into Blender cubes cubealign cube2raw cubegen coords raw blender

59 Getting Input Gaussian/ cubegen cubes Quantum Data into Blender cubes cubealign cube2raw cubegen coords raw blender

60 Getting Input Gaussian/ cubegen cubes Quantum Data into Blender cubes cubealign cube2raw cubegen coords raw blender images

61 Negative Oxygens Almost Neutralized Oxygens Positive Calcium Amine e-potential in Atomic Units And we can actually make a movie how charge and electron density change as the calcium ion moves away

62 Everything Assembled Together! The Thrombospondin Movie

63

64 Reims: Thanks to my French partners: CNRS UMR 6237 Manuel Dauchez Laurent Martiny Paris: INSERM UMRS 665 Catherine Etchebest CNRS UMR 6229 Eric Henon The Champagne Ardenne Computational Center Romeo The Champagne Ardenne region for funding my work