The Unix Cray station is now installed on the Silicon Graphics Iris 3130 workstation at the Atlas Centre, using the NSC Hyperchannel and is being made available to users.
An Interpreter for CGM metafiles has been written so that metafiles written on the Cray using GKS can be displayed locally. This service is intended to make use of the high speed link to the Cray for viewing results quickly in order to control programs, but not for full interaction. The advantage of this form of working is that it is possible to change program parameters after seeing partial results. Also the graphics produced are of a very high quality.
Other forms of viewing and interactive Graphics are being developed that utilise the full potential of this workstation and its link to the Cray. This includes software to produce fully rendered images of 3D objects. It is also hoped to buy a high quality colour printer that can produce faithful reproductions of the screen image.
Existing software that runs on the Iris is also being evaluated, but the choice of software made available depends on user demand. If you have any special software requirements, RAL Graphics Group would like to hear from you.
Anyone wishing to use this service should contact Roy Platon in RAL Graphics Group (RTP(@UK.AC.RL.IB) for further details.
Electron transfer reactions play a fundamental role in all photosynthetic and respiratory processes in living organisms. These processes are controlled chains of reactions in which the redox potentials of the individual steps are finely balanced. Many of these reactions involve electron donors/acceptors in which metal atoms are firmly attached to proteins (molecules which contain several thousand atoms and have an irregular structure). The electrons often transfer over distances in excess of 15 Angstroms, and frequently have to pass through the protein en route. The protein has a profound influence on the reaction, and provides a fixed, heterogeneous environment in which the electron transfer takes place. Due to the complexity of the system, despite extensive experimental and theoretical work a lack of understanding persists regarding the fine details of control of the energetics, and also of the motion of the electron through the system.
The reactions are very difficult to model. Standard quantum chemical procedures cannot cope with the size of the system. Current molecular mechanics programs for proteins allow for simple ('classical') modelling of the system, but cannot represent the unavoidable quantum mechanical details of the electron. In recent years, a new technique has been developed called path integral Monte Carlo (PIMC), which is based on a formulation of quantum mechanics derived by Feynman. It enables the simulation of a system with a mixed 'classical' and 'quantum' representation in a fully consistent manner. This is ideal here, as only a small part of the system is expected to display sizable quantum effects on the scale of the reaction. A 'quantum particle' is represented as a string of 'beads', each of which can be treated during the simulation in a similar manner to the 'classical particles'. Thus, PIMC provides a prescription for converting a quantum mechanical problem into one routinely dealt with in standard molecular mechanics.
Some major problems remain to be overcome in using PIMC to study protein electron transfer reactions. Firstly, new potentials are required to describe the interaction of the electron and the protein. Because of a lack of a numerically stable algorithm for simulating more than one electron explicitly in PIMC, these potentials must account for some of the electron-electron interactions, as only the transferring electron can be explicitly represented. Secondly, for similar reasons, at present dynamics cannot be simulated for this system using PIMC.
Work is currently in progress to derive and test improved potential models for the electron-protein interaction. Presently, benzene is being used to generate and test models of a pi-electron-containing system. After success with benzene, it is intended to initiate a study of the redox protein cytochrome ct, in which intramolecular electron transfer has been reported and for which an accurate structure has been determined. From initial studies, this work is expected to require around 100 hours of CPU time per protein simulation.
Hopefully, this work will indicate the region of the protein through which the electron travels, and also the importance of individual chemical groups in contributing to the electron-protein interaction, and hence to the overall influence of the protein on the reaction.
