Modeling DNA with Graphite-MicroMegas Samuel Hornus Damien Larivi` - - PowerPoint PPT Presentation

modeling dna with graphite micromegas
SMART_READER_LITE
LIVE PREVIEW

Modeling DNA with Graphite-MicroMegas Samuel Hornus Damien Larivi` - - PowerPoint PPT Presentation

Modeling DNA with Graphite-MicroMegas Samuel Hornus Damien Larivi` ere INRIA Fourmentin-Guilbert Foundation A partnership to tackle the big challenges of biological 3D modeling Agenda Biological data deluge Data mass and


slide-1
SLIDE 1

Modeling DNA with Graphite-MicroMegas

Samuel Hornus Damien Larivi` ere INRIA Fourmentin-Guilbert Foundation

A partnership to tackle the big challenges of biological 3D modeling

slide-2
SLIDE 2

Agenda

➲ Biological data deluge ➲ Data mass and comprehension ➲ Comprehension by 3D modeling ➲ Modeling DNA is required ➲ MicroMégas is of great help

slide-3
SLIDE 3

Biological data deluge

➲ Modern sequencers: One human genome every 14

minutes, 1-2 TB of raw data

A, T, G, C,...

Stuart M. Brown, NYU Langone Medical Center

slide-4
SLIDE 4

Biological data deluge

➲ Fluorescence

microscopy:

  • Basis of

numerous experiments

  • Acquisition of

millions images per run

  • Generates

several tens

  • f terabytes

Pepperkok & Ellenberg, Nature, 2006

slide-5
SLIDE 5

Biological data deluge

Julio.Ortiz, Max Planck Institute of Biochemistry

Electron microscopy

slide-6
SLIDE 6

Data mass and comprehension

➲ 50 years used to dissect cells ➲ Time is come to re- assemble the disconnected

parts

2 µm ~ 3 millions components at the right place

http://www.asiatraveltips.com/newspics/0611/AirbusA380inHK2.jpg David Goodsell, TSRI

slide-7
SLIDE 7

Comprehension by 3D modeling

➲ 3D modeling is being adopted ➲ 3D software borrowed from “Hollywood” ➲ Need a career time to be learned

Entry of Dengue virus in a cell (Janet Iwasa)

slide-8
SLIDE 8

Modeling DNA is required

Mycoplasma cell (David Goodsell)

slide-9
SLIDE 9

Modeling DNA is required

➲ Standard molecular tools and webservices:

  • Lack of intuitivity
  • Very basic modeling functions

➲ The opensource and intuitive MicroMégas plugin

  • vercomes this situation
slide-10
SLIDE 10

MicroMégas is of great help

The bacterial DNA repair system

Winkler et al, J. Biological Chemistry, March 2011

slide-11
SLIDE 11

Geometry of DNA

Naive view of DNA is good for low-level modeling:

  • string-like structure
  • helical shape
  • long sequence of very similar “base pairs” (ACGT)

[google image]

slide-12
SLIDE 12

Geometry of DNA

Computer graphicists translate this structure to:

  • a curve
slide-13
SLIDE 13

Geometry of DNA

Computer graphicists translate this structure to:

  • a curve
  • a uniform sampling of orthonormal frames
slide-14
SLIDE 14

Geometry of DNA

Computer graphicists translate this structure to:

  • a curve
  • a uniform sampling of orthonormal frames
  • instancing of base pairs
slide-15
SLIDE 15

Geometry of DNA

Computer graphicists translate this structure to:

  • a curve
  • a uniform sampling of orthonormal frames
  • instancing of base pairs
  • with twisting: rotation around the tangent vector
slide-16
SLIDE 16

Modeling a curve

Standard curve models:

  • Quadratic B´

ezier curve

  • Cubic B´

ezier curve

  • Special case when input is a bare sequence of points

= ⇒ interpolatory subdivision scheme [Dyn, Floater and Hormann 2009]

Modeling a curve

slide-17
SLIDE 17

Uniform sampling

Generating a uniform sampling with tangent is easy Generating a normal at each sample point is difficult

slide-18
SLIDE 18

Uniform sampling

Generating a uniform sampling with tangent is easy Generating a normal at each sample point is difficult We want a continuous frame that minimizes torsion E.g. the Fr´ enet-Serret frame is not continuous

slide-19
SLIDE 19

Uniform sampling

Generating a uniform sampling with tangent is easy Generating a normal at each sample point is difficult We want a continuous frame that minimizes torsion E.g. the Fr´ enet-Serret frame is not continuous Recent technique: very fast and very good approximation: [Rotation Minimizing Frames, ACM ToG 27(1):2, 2008]

slide-20
SLIDE 20

Visualization with instancing and “ray-casting”

Use OpenGL to instantiate a 3D model of a base-pair in each frame along the curve:

  • Setup GL transform matrix
  • One call to glDrawArrays to draw one base pair
slide-21
SLIDE 21

Visualization with instancing and “ray-casting”

A base-pair has ≈ 40 atoms. We setup GLSL programs so that:

  • input = array of atoms {center, radius, color}
  • geometry shader builds a quad in front of the atom
  • pixel shader compute intersection of ray & atom (a sphere)

Use OpenGL to instantiate a 3D model of a base-pair in each frame along the curve:

  • Setup GL transform matrix
  • One call to glDrawArrays to draw one base pair
slide-22
SLIDE 22

Visualization with instancing and “ray-casting”

A base-pair has ≈ 40 atoms. We setup GLSL programs so that:

  • input = array of atoms {center, radius, color}
  • geometry shader builds a quad in front of the atom
  • pixel shader compute intersection of ray & atom (a sphere)

Use OpenGL to instantiate a 3D model of a base-pair in each frame along the curve:

  • Setup GL transform matrix
  • One call to glDrawArrays to draw one base pair

atom camera

slide-23
SLIDE 23

Visualization with instancing and “ray-casting”

A base-pair has ≈ 40 atoms. We setup GLSL programs so that:

  • input = array of atoms {center, radius, color}
  • geometry shader builds a quad in front of the atom
  • pixel shader compute intersection of ray & atom (a sphere)

Use OpenGL to instantiate a 3D model of a base-pair in each frame along the curve:

  • Setup GL transform matrix
  • One call to glDrawArrays to draw one base pair

atom camera

slide-24
SLIDE 24

Level of Details

slide-25
SLIDE 25

Level of Details

slide-26
SLIDE 26

Level of Details

slide-27
SLIDE 27

Level of Details

slide-28
SLIDE 28

Level of Details

Hierarchy used for

  • base-pair picking (on mouseclick)
  • LoD selection w.r.t camera position
slide-29
SLIDE 29

Level of Details

← − Camera is left of screen

slide-30
SLIDE 30

Thank you

[demo?]

slide-31
SLIDE 31

Why modeling biological scenes spatially?

Medical illustration

  • Popularizing knowledge
  • Help scientists’ understanding
  • Dynamic simulation

Mre11 David Goodsell c 2010

slide-32
SLIDE 32

Why modeling biological scenes spatially?

Medical illustration

  • Popularizing knowledge
  • Help scientists’ understanding
  • Dynamic simulation

Mre11 David Goodsell c 2010

Scientific reasons specific to DNA

  • Later in this talk...
slide-33
SLIDE 33

Modeling DNA

Growing importance of modeling and simulation for experiments = ⇒ need for specialized spatial modeling tools for biologists = ⇒ DNA is an important target

slide-34
SLIDE 34

Modeling DNA

“Modeling DNA in space is such a tedious job!” — microbiologists and illustrators

slide-35
SLIDE 35

Modeling DNA

“Modeling DNA in space is such a tedious job!” — microbiologists and illustrators Use advanced 3D modeling software

  • Maya • 3D Studio • Blender

Or command-line tool and web services with form-based input

  • 3DNA • 3D-DART • DNA Maker

And some with “UI”

  • VDNA (VMD plugin)
slide-36
SLIDE 36

Modeling DNA

“Modeling DNA in space is such a tedious job!” — microbiologists and illustrators Use advanced 3D modeling software

  • Maya • 3D Studio • Blender

Or command-line tool and web services with form-based input

  • 3DNA • 3D-DART • DNA Maker

And some with “UI”

  • VDNA (VMD plugin)