Issues Surrounding the Use of Open-Source PyMOL to

Visualize Large Complexes

Warren L. DeLano, Ph.D.

DeLano Scientific LLC

Overview

1. Project Overview

2. Some “Biggish” examples

3. Animations

4. Intro/Demo (if there’s time)

What is DeLano Scientific LLC?

• LLC = Limited Liability Company

• Mission: to create, share, and support molecular software for all research scientists, students, and educators, with minimal restrictions on usage.

• DeLano Scientific can be thought of as:– A private software company with an independent

quasi-academic, scientific focus. or

– An independent quasi-academic laboratory funded as a private scientific software company.

Open-Source vs.Free for Academics

• PyMOL is true Open-source in that it does not discriminate against commercial entities – it is free for all for unrestricted reuse.

• However, extra benefits are provided to those individuals, laboratories, schools, and companies who voluntarily sponsor the effort with their contributions.

How are the Bills Paid?• Sustainable funding now derived from purchases

of PyMOL licenses and support agreements.

• Therefore, we are directly accountable to hundreds (eventually thousands?) of users in Academia and Industry for our continued existence and growth.

• To survive as a business, we must rapidly come up with practical solutions to broad visualization needs for today and the future.

Visualization of Large Complexes

• Software Issues– What can you do with PyMOL?– Where does PyMOL break down?– What can be done about it?

• Data Format Issues– Existing limitations in the data files.– PyMOL is a mostly a follower/consumer in this

area.

PyMOL Is a General Tool

• PyMOL is good for many things, but not great for very many of them.

• User base spans the full range: students, small molecule chemists, crystallographers, up through CryoEM.

• Best uses of PyMOL at present:– Communication of “molecular scenes”.– Generation of high quality images.– Scripting and generation of animations.

Visualization of Large Complexes

• Can combine molecular data with volumetric data.

• Can compute volumes from atomic coordinates.

• Can compute isosurfaces and color-by-potential (no “volume” rendering yet)

Can PyMOL Visualize Large Complexes?

• Well…maybe.

Let’s try some…

Limits of PyMOL

• There are some practical limits…– Maximum = 1 million atoms on a 1 GB machine.– The comfortable limit on Linux is about a 1/4 of that.– The practical limit on Windows is only about 150,000

atoms per GB of RAM due to VM problems.

• PyMOL is optimized for performance and image quality, but not memory usage.

• Unless specifically prioritized by users, memory usage is not likely to improve much due to architectural constraints in the code.

• However, it may be possible to improve Windows memory behavior to approach that of Unix.

• 8+ GB RAM should become attainable over the next several years as 64-bit computers become more common.

• G5, Athlon64, Itanium2, etc.

Outlook for Improvement?

Map File Formats:

• PyMOL does has volume data support:

– Electronic density• XPLOR, CCP4, O/Brix

– EM reconstructions• No specific file formats yet

– Electrostatic Potentials• Delphi/Grasp, Biosym Grids, Mead AVS Grids

Limits on PyMOL Map Handling

• Fast but wasteful– PyMOL currently stores vertices and levels– 4 * sizeof(float) = 16 bytes/voxel– Largest practical map with 1 GB RAM

200 ^ 3 ?

• Current user base hasn’t yet bumped up against PyMOL’s limits – but it may happen.

Summary on Large Complexes

• PyMOL may be useful for working with asymmetric units.

• However, more specialized tools will probably be needed for visualization of million-atom biological-units.

Animations

1) Rendered Movies

2) Real Time OpenGL Movies

Real Time Animations

Demo