hiviz : visualizing an hiv envelope protein

HIViz: Visualizing an HIV Envelope Protein

Philip HellerCMPS 261

Project PresentationSpring 2010

Motivation

• 33M people living with HIV/AIDS– 2M children– 22M in sub-Saharan Africa

• 25 years of vaccine research– Traditional biomedical approaches– Not much visualization– Recent success in Thailand energizes efforts

www.dicid.org

Wikimedia.org

www.wayfaring.info/category/topics/museums/page/9

“The smallpox, so fatal, and so general amongst us, is here

entirely harmless …”

“The old woman comes with a nut-shell full of the matter of the best sort of small-pox, and asks what vein you please to have opened. She immediately rips open that you offer to her, with a large needle … and puts into the vein as much matter as can lie upon the head of her needle …”

Letter from Lady Mary Wortley Montague to Mrs. S. C.1719 (?)

“I intend to try it on my dear little son.”

Preventable by Vaccine

• Chickenpox• Diphtheria• Hepatitis A• Hepatitis B• HPV• Flu• Measles

• Mumps• Pertussis• Rubella• Shingles• Tetanus• Polio• ?????HIV?????

FamousPolioSurvivors

www.observer.com

www.setileague.org

www.tvworthwatching.com

www.lyricsystem.com

Why HIV is Different

• No one has ever recovered• HIV destroys immune cells • HIV inserts genome into host cell genome,

hidden from immune system• HIV evolves constantly/rapidly

Many Vaccine Efforts Focus on gp120

• Envelope surface• Entry into host cell• Crystallized 1998

Berman Lab Data:Primary Sequence Loci of Sites for…

• Protease binding (% prevalence)–Possible cutting site

• Receptor binding (absent/present)–1st contact with host cell

• Glycosylation (absent/present)– Thick carbohydrate coat, hard to attack

• Neutralizing antibodies (absent/present)• Positive selection (all mutations)

Positive Selection

• Infection begins with a “founder” individual virus• High mutation rate, rapid reproduction => many

variants• Mutations conferring positive or neutral traits are

propagated• Selective sweeps• SNP sites are identified, qualified, not quantified

gp120 3D Viz: SOTA

• All derived from Wyatt & Kwong 1998 crystallization

• Converted to PDB file format– 3D location of all atoms

• RasMol, PyMol, JMol for viz & markup• Markup script supplied at:– Startup time (file)– Run time (keyboard)

media.caltech.edu

www.pdb.org

WHAT IF …

• What if I could see all features in 3D?• Annotate image with feature observations• Visualize with modified JMol• Goals:– Simultaneously view all 5 features– Predicates eg “Glycosylation –OR– Neutralizing Abs”– Software generation & application of scripts– Extensible: other proteins/features

Simultaneous Viewing of all 5 Features

• Too much for 1 image• 5 images: – +: Looks good, easy, supports 6th “predicate” view– -: Very hard to interpret multiple orientation

Solution: “Synchronized Swimming”

• Echo mouse events in any window to all others• All zoom/roll/pitch/yaw commands issued

identically to all instances• (Probably) novel– JMol not designed for SC/MV– Requires some source-level understanding of Jmol

• Glance from image to image without mentally re-orienting

Ad-Hoc Boolean Queries

• What sites support protease binding sites are also glycosylation sites?

• What sort of mutation happens at antibody neutralization sites?

• Predicate-logic combinations of features

A 6th “Predicate” window, + config dialog

Predicate Logic on Numbers• Protease binding: % prevalence• Mutation: What amino acids have been seen at each locus (up to

20, but that’s rare)• Others: Absent/Present• Convert to [0.0 – 1.0]

– Prevalence / 100– Mutation: (# of different amino acids) / 20– Absent = 0, Present = 1

• Numerical boolean operations:– A or B = max(A,B)– A and B = min(A,B)– Not A = 1 - A Demo Time

Conclusions

• Multiple views are a good compromise between Single-View-Busy-Features and Single-Feature-Busy-Views

• Ad-hoc predicate view supports scientific inquiry