Is that not a little bit splitting hairs? If you visit the Princess Margaret Hospital website and everything around it, it breaths nothing but cancer involvement. The "Help Conquer Cancer" credo has long long existed. The HCC product is primarily aimed at better / faster cancer diagnosis, crucial, but the good thing is that the knowledge gained will benefit many other medical research. Any medical research project brings x-disciplinary information to surface, so think, with Cancer research funding behind, it's A Okay "Total Irre" with me.

Ah ja, the main researches link states what he does and the owners are determining factor of a project name, not WCG! http://www.cs.toronto.edu/~juris/

cheers

Michael,

Since I am motivated as much by a desire to understand the "nuts and bolts," of scientific research as by some abstract desire to increase scientific knowledge, I am also annoyed by these promotional gimmicks. They tend to confuse this retired lab tech. Heck, I only have a technical degree.

When I studied biochemistry, back in the 70's, I was told that proteins were freeze dried into crystals, and then x-rayed. This liquid crystal stuff is new to me. Not that I understood anything about x-raying freeze dried protein crystals either!

I would like to know what they mean by wanting to "improve the CrystalVision system to enable automated, accurate and fast crystallography image classification."

I googled CrystalVision AND crystallography, and Google returned, among other things, a trademarked protein analysis system. Since the University Health Network does not indicate that CrystalVision is trademarked, it probably is not the same thing.

============
Does anyone know of a distributed computing team that consists of people with only one average computer that they turn off at night? That's about the only type of distributed computing team that I would join. I like to talk about WHAT I am crunching more than HOW FAST I am crunching.

===========

Jack

Jack, that could be the start of an interesting discussion but in this thread that would be typically off topic. Therefore if you actually like to share views on this subject you should post your question in a new thread in the Chat Room section.

See you there?. Jean.

Michael,

Proteins consist of hundreds to thousands of atoms. Incompletley cryatalized proteins can give you parts of the structure. Enough partial images can give you a hint as to what conditions (Ph, solvent, vibration isolation, etc.) will give a better crystal.

The project applies multiple image analysis algorithms to a set of over 100K images of crystallization experiments. Each of these experiments has been scored visually by multiple researchers and provides a gold standard training set including all the typical outcomes. The aim is to identify which or which combination of image analysis algorithms provide best agreement with the human classified images. Those algorithms will then be used to automatically classify >9 million other previous experiments and new experiments as they are conducted in realtime. With something like 9000 different macromolecules, many related to cancer, automated scoring of all the results along with the incomplete factorial sampling of chemical space may provide predictive techniques to crystallize samples. It may for example, be possible to identify subgroups of macromolecues that respond better to crystallization from crowding, e.g. with PEGs, or those that like conditions with salt in. Invariably the more important the sample, then the harder it is to get. A little predictive knowledge can go a long way in optimizing the techniques applied to the sample. Crystallization is still an empirical process as it is very difficult to predict how a biological macromolecule will crystallize without knowing it's three-dimensional structure (obtained from X-raying the crystal). Even knowing the structure it's hard to predict other crystallization conditions.