1

Beyond BallBeyond Ball --andand --StickStickPart 2: Practical Chemistry VisualizationPart 2: Practical Chemistry Visualization

Mario ValleMario ValleSwiss National Supercomputing Centre (CSCS)Swiss National Supercomputing Centre (CSCS)

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

What we have coveredWhat we have covered

1. Visualization fundamentals

2. Basic chemistry data types and representations

3. Some ideas about breaking barriers and increase usefulness of visualization

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

What we cover nowWhat we cover now

1. Data and data management

2. Good visualization tool characteristics

3. Visualization tools here at CSCS

4. Digital storytelling tools and ideas

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Data issuesData issues

� Representation follows data logical structure andintended usage

� Horror stories from not knowing the data

� One thing is logical data format, another the physical data file format

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Usual chemistry data typesUsual chemistry data types

O

O

O

OH

Data from prof. A. Oganov – ETH Zürich

Structures

Scalar volumes

1D and 2D tables

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Non quantitative dataNon quantitative data

2

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

StructureStructure

� Atoms coordinates

� Atom types

� Eventually bonds data

� Optional scalar values (like charges) or vector values (like vibration modes)

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Structure visualization goalsStructure visualization goals

� Show spatial configurations

� With time show peculiar movements

� Show correlation between position/structure and other quantities

� Show matching or spatial related configurations

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Basic structure representationBasic structure representation

Ball and Stick

Licorice

CPK

Surfaces

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

High level structuresHigh level structures

Accessible surface Secondary structure Secondary structure

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Less is moreLess is more

Data from prof. A. Oganov – ETH Zürich

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Problem: structures are too bigProblem: structures are too big

3

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Related scalar quantitiesRelated scalar quantities

Colored by atom type Colored by charge

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Vector data (vibration modes)Vector data (vibration modes)

� Static (arrows)

� Animated

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Visualization techniquesVisualization techniques

� Select part of a structure

� Show two or more structures together

� Show surfaces

� Add high level “summary” geometries (polyhedrons, planes, etc.)

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Data formatsData formats

Base for almost everything

� PDB

� Gaussian Cube

� Etc.

But there are incomplete formats (missing atoms types,

unit cell, etc.)

� Other?

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Time dependent dataTime dependent data

Frame-by-frame

Summary over trajectory

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

FrameFrame --byby --frame or summary?frame or summary?

Images from AmiraMol

Sergey Churakov – PSI Villigen

4

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

TimeTime --dependent visualization goalsdependent visualization goals

� Show spatial configurations changing over time

� Show phase transitions

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Trajectory data formatsTrajectory data formats

� Kino

� PDB

� List of PDB

� DCD

Working around limitations: Kino + file with base vectors

� Other ?

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Volume dataVolume data

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Volume data file formatsVolume data file formats

The following formats contains structure + uniform grid

of scalar values:

� Gaussian Cube

� CHGCAR

� Other?

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

1D data1D data

Davide Donadio – ETH ZürichBeyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Choose the right mappingChoose the right mapping

25

27

29

31

33

0 3 6 9 12 15

time

tem

p

25

27

2931

32

32

ABC analyzer warm-up

25

27

29

31

33

0 3 6 9 12 15 18

time from power-on (min)

tem

pera

ture

(°C

)

5

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

2D Tables2D Tables

COSY NMR Spectra

Trajectory in parameter space

Dotplotfor protein

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

3D time dependent data3D time dependent data

Added perceptual cues to help understand 3D trajectory in parameter space

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

1D & 2D tools1D & 2D tools

� Gnuplot

� Grace

� Scigraphica

� R

� Matlab

� Other?

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Trend in scientific data lifecycleTrend in scientific data lifecycle

From this (publish and forget)…

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Trend in scientific data lifecycleTrend in scientific data lifecycle

…to this (use, reuse, recycle)…

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Metadata are not evil!Metadata are not evil!

Metadata is information about data

We often use metadata without even knowing it

If you had two cans without labels, which would you eat?

Without a label, how would you know which was tuna and which was cat food?

Cat Food?

Tuna?

6

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Metadata inside/outside dataMetadata inside/outside data

� Inside the file• E.g. PDB

� TITLE (describe what the PDB is about)� KEYWDS (some keywords to retrieve the file)

� AUTHOR (who to blame for bad data)� REMARK 6 – 99 (free form remarks)

� Encoded in the filename or file path/simulations/20050921/param1=0.033/result_run_1.pdb

� In a companion file• Protein.pdb + Protein.xml• Desc.xml for a set of files

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Multiple files, different parametersMultiple files, different parameters

Noise variance = 0

0.003

0.010

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

My chemistry file collectionMy chemistry file collection

Description file editor

Generated pageview in browser

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

The quest for the perfect toolThe quest for the perfect tool

The perfect tool makes everyone happy. But:

� Chemistry is a very wide design space

� No single definition of perfection

� Visualization is still an art

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

The visualization “Holy Grail” The visualization “Holy Grail”

Smartsystem

Your data

Perfectvisualizations!

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Perfect tools Perfect tools –– somewhere elsesomewhere else

7

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Chemistry visualization tools Chemistry visualization tools

What a chemistry visualization tool should provide:

� Load and display everything I’m working on

� Enable exploration and comparison

� Produce high quality images and movies for publication

� Has flexibility in adding customizing visualization techniques or analysis scripts

� Let me experiment with techniques and rendering modes

But there are difficulties…

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Difficulty 1: tools inflexibilityDifficulty 1: tools inflexibility

Nice crystallography programs, but animation missing and no way to extend them

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Difficulty 2: formats a Difficulty 2: formats a goggog òò

PDB, Gaussian Log, Gaussian Cube, Plan coordinates,PDB, Gaussian Log, Gaussian Cube, Plan coordinates,Kino, SHELKino, SHEL --X, VASP POSCAR, XDATCAR,X, VASP POSCAR, XDATCAR,

ADF, DCD, DL_POLY, VASP XDATCAR,ADF, DCD, DL_POLY, VASP XDATCAR,Concatenated VASP POSCARConcatenated VASP POSCAR

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Difficulty 3: minimal integrationDifficulty 3: minimal integration

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

ChemViz@CSCSChemViz@CSCS

� We do not endorse any specific program, but encourage you to use the tool best suited to you research

� We use the STM3 platform to implement unusual and advanced techniques

� The other two application we suggest (if you ask) are:� VMD� Molekel

� The OpenBabel file type converter could be useful for accessing strange file formats

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

STM3 toolkitSTM3 toolkit

STM3 is a framework in which to develop unusualand enhanced techniques for molecular visualization

STM3 goal is not to supplant existing tools

The toolkit is built on top of the commercial visualization environment AVS/Express

8

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

STM3 modulesSTM3 modules

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

VMDVMD

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

(new) (new) MolekelMolekel

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Tools resourcesTools resources

� STM3� http://www.cscs.ch/~mvalle/ChemViz/� http://www.cscs.ch/projects/AVSChemistry.php

� VMD� http://www.ks.uiuc.edu/Research/vmd/� http://www.theochem.ruhr-uni-bochum.de/~axel.kohlmeyer/cpmd-vmd/

� Molekel� http://www.cscs.ch/a-display.php?id=138

� OpenBabel� http://openbabel.sourceforge.net/

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Digital storytellingDigital storytelling

We “see” a good story develops in our mindBeyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Different roles of visualization Different roles of visualization

2. Communicate and present results

1. Help understanding datawww.smartmoney.com/marketmap

www.peets.com/selector_coffee/coffee_selector.asp

9

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Why scientists are often boring?Why scientists are often boring?

“...drawing graphs, like motor-car driving and love-making, is one of those activities which almost every researcher thinks he or she can do well without instruction.”

Wainer & Thissen, 1991Annual Review of Psychology

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

A not so unusual presentation…A not so unusual presentation…

When we added no drug the viral infection spread and lead to cell death within 17 hours of the initial challenge.

When we added the drug at 5 nM the viral infection persisted andspread in the culture leading to cell death within 18 hours postchallenge.

When we added the drug at 5 nM levels the 78% of cells survived viral infection and grew slowly without cell division.

When we added the drug at 10 nM levels 73% of cells survived viral infection and grew slowly without cell division.

When we added the drug at 15 nM levels 6 % of the cells resistedviral infection but did not grow or divide.

When we added the drug at >15 nM levels all of the cells died within 10 hours.

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

…… that could be betterthat could be better

-/-

-/-

+/-

+/+

+/+

Growth/division

100%10 hrs> 15 nM4

94%∞15 nM3

∞

18 hrs

17 hrs

Survival time post challenge

27%10 nM2

22% 5 nM1

0%0control

Toxicity[Drug]

Protection from viral infection

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Excellence in thinkingExcellence in thinking

“[…] clarity and excellence in thinking is very much like clarity and excellence in the display of data. When principles of design replicate principles of thought, the act of arranging information becomes an act of insight.”

Edward Tufte 1998 p. 9

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Visual messages capture attentionVisual messages capture attention

PowerPoint 1.0 derives from a product called “Presenter” developed by Forethought Inc. at the beginning of 1987.

Microsoft bought Presenter on august 1987 for 14 millions of dollars.

Immagine da: albinoblacksheep.com

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Visual preempts verbal messagesVisual preempts verbal messages

100

101

102

103

104

105

1999 2000

A verbal message never dominates over a non verbal one.

We walk away from this chart thinking that 2000 results are twice the ones from 1999 (instead of a mere +2%).

10

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Visual messages capture attentionVisual messages capture attention

Do you remember how much Microsoft paid Presenter?

Do you remember what the guy was doing in the photo?

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

““““14 Ways to Say Nothing with Scientific Visualization”14 Ways to Say Nothing with Scientific Visualization”14 Ways to Say Nothing with Scientific Visualization”14 Ways to Say Nothing with Scientific Visualization”Al Al Al Al GlobusGlobusGlobusGlobus, Eric , Eric , Eric , Eric RaibleRaibleRaibleRaible –––– NASA NASA NASA NASA –––– July 1994July 1994July 1994July 1994

14 Ways To Say Nothing…14 Ways To Say Nothing…

Right!Right!Right!Right! Wrong (violates rules)Wrong (violates rules)Wrong (violates rules)Wrong (violates rules)

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

14 Ways To Say Nothing…14 Ways To Say Nothing…

1. Never Include a Color Legend

2. Avoid Annotation

3. Never Mention Error Characteristics

4. When in Doubt, Smooth

5. Avoid Providing Performance Data

6. Quietly Use Stop-Frame Video Techniques

7. Never Learn Anything About the Data or Scientific Discipline

8. Never Compare Your Results with Other Visualization Techniques

9. Avoid Visualization Systems (e.g. AVS)

10.Never Cite References for the Data

11.Claim Generality but Show Results from a Single Data Set

12.Use Viewing Angle to Hide Blemishes

13. If Viewing Angle Fails, Try Specularity or Shadows

14. “This is easily extended to 3-D”

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

SiggraphSiggraph 1993: VIZ1993: VIZ--OO--MATICMATIC

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Why chemistry should be boring?Why chemistry should be boring?

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

It is not a media, but it is importantIt is not a media, but it is important

Design your presentation remembering that not all your readers are perfect:

There are color blind people

� Don’t rely on color alone for decisions

� Use appropriate color schemes

There are people with low sight

� Avoid low contrast in visualizations and icons

� Avoid tiny fonts

11

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Media and toolsMedia and tools

� Images� Usual problems� Tools

� Movies� Usual problems� Tools

� Web

� New, unusual output platforms?

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Images Images –– usual concernsusual concerns

� Image file formats� No JPEG, better using PNG� Leave TIFF for process printing� Images for the web: JPEG, PNG, Animated GIF

� Quality� Antialiasing lines and borders

� Resolution� Screen 75dpi � better than 1000dpi for printers

� Colors� Unreadable colors (especially on conference beamers!)� Bad colors in print vs. good on screen� Black & White requested by some publications

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Image formatsImage formats

PNG

JPEG

“mosquitoes” aroundsharp edges

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Image qualityImage quality

Edges on printed paper looks terrible at the printer resolution

Screen Paper

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Color gamutColor gamut

Each media has aset of colors that can be reproduced

This set is called the device color gamut

The number of colors available quantize the color gamut. If the set is too small artifacts (color banding) arise

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Swiss Knife for imagesSwiss Knife for images

Image Magick <www.imagemagick.org>

convert img-from img-toconvert –scale 50% img-from img-toconvert –flip img-from img-toconvert img-from –colorspace Gray img-toconvert img-from –annotate 0x0+50+10 ‘© Mario Valle 2005’ img-toconvert img-from –draw ‘image Over 50,10 0,0 logo.tif’ img-to

montage –tile 2x2 –geometry 512x512 img[1234].png tot.png

identify –verbose image.jpg

12

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Other useful image toolsOther useful image tools

� GIMP <www.gimp.org>

� Littlecms <www.littlecms.com>

� Photoshop

Check <www.cscs.ch/~mvalle/postprocessing/>

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

LittleCMSLittleCMS example of processingexample of processing

Original image

Color corrected for HP1200PS

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Different rendering intentsDifferent rendering intents

Rendering for visualization Rendering for presentation

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Movies Movies –– usual concernsusual concerns

Codec availability� Impact playability on different machines. Unfortunately there

is no universal no-fuss coding method� Codec and container file format are two distinct things

Quality� Depends strongly on bitrate and compression method

(trades compression time for quality)� Suggestions for natural movies did not apply to sharp edge

‘cartoon-like’ movies

File size� Related to bitrate (quality), frame size, compressibility

Output usage� Personal projection, web, TV program, DVD burning, etc.

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Input to movie creationInput to movie creation

� Usually a set of frames (image files)

� Better if the images are uncompressed (not to lose quality)

� Should choose a framerate (usually 10-12 fps, but some standards have a fixed framerate)

� Not too much frames! (500 frames @ 12 fps � 42 sec)

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

CodecsCodecs and containersand containers

Normally everything is perfect on your own workstation......and nothing works on the conference room PC.

Codecs� MPEG1� MPEG4 (different implementations: MS Mpeg4 V2, Divx, xvid)� X264

Containers� AVI� MOV� MPEG� MP4

13

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Movie toolsMovie tools

Mencoder� My preferred workhorse� It has a myriad of options (difficult to master)

Adobe Premiere� Everything you can dream of� Perfect for adding titles and transitions

VirtualDub� Simple editing

Other� MJPEG, transcode, ffmpeg, ffmpegX, etc.

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Movie playersMovie players

Default ones� Windows Media Player� Quick Time Player (a triumph of non-usability)

Other ones� Xine (plays AVI on Linux)� Mplayer (plays almost everything on Linux and Windows)� VLC (quick and multiplatform)

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

GUI for GUI for mencodermencoder

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Web pageWeb page

� Has unique opportunity for more interactive experience. Applets can be used in place of static images.

� The browser is an universal user interface for data collections and some decision support systems.

� Limits are in the available screen space and the heterogeneity of the client browsers (for platform, type and installed plugins)

� Load time is a critical factor, if it is too long, the user is discouraged and can go away.

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Consider for web displayConsider for web display

Movies� Download time� Provide more than one resolution

Images� Reduce number of colors� Reduce size

Applets� Chime

Animated GIF� Quick and light

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Unusual media: PDA & Unusual media: PDA & cellphonescellphones

Currently very

limited and not an usual medium, but who know...

14

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

Immersive visualizationImmersive visualization

� Enhanced interaction experience: you are “inside” your data.

� Needs special hardware to view, interact and trackuser position

� Besides stereo projection it addsviewpoint change

Beyond Ball-and-Stick Tutorial – Mario Valle – CSCS User Day 27/09/2005

What we have coveredWhat we have covered

1. Data and data management

� Know your data

� Record useful metadata

2. Good visualization tool characteristics

� Do not search the perfect tool, but think about your goal

3. Visualization tools here at CSCS

� Ask and consult with us!

4. Digital storytelling tools and ideas

� Present to persuade and communicate

Beyond BallBeyond Ball --andand --StickStick

Thanks for your attention!Thanks for your attention!

Mario ValleMario Valle

mvalle@cscs.chmvalle@cscs.chhttp://www.cscs.ch/~mvalle/ http://www.cscs.ch/~mvalle/