Technology : Versatile liver inspires parallel processing

时间:2019-02-27 10:15:01166网络整理admin

By Mark Ward THE human brain is second to none when it comes to processing power. But computer scientists are now investigating the liver as a model for handling information, because it is so adaptable and carries out so many different tasks simultaneously. Researchers at the University of Liverpool are creating a computational version of the liver which they hope will be the inspiration for more efficient parallel processors, as well as adaptive computer systems. The basic building blocks of the liver are hepatocytes. These cells perform various tasks: secreting bile, stripping toxins out of blood, metabolising proteins, carbohydrates and fats, regulating blood volume and making clotting agents. Hepatocytes are organised into hexagonal structures called lobules, in which flat plates of hepatocytes radiate from a central vein to the outer wall of the lobule. Blood flows through gaps called sinusoids on both sides of the plates and is processed by the hepatocytes. The processed blood then passes into the central vein. Bile produced by the hepatocytes passes to the outer wall of the lobule to exit via ducts. The liver has more than a million lobules, but not all of them do the same job at the same time—they switch tasks depending on what challenges the blood presents them with. This, in turn, depends on whether the owner of the liver is drinking, dining or has a disease. Ray Paton of Liverpool’s computer science department says it is this multi-level, adaptable configuration that he and his colleagues are trying to emulate in their computers, but in their case they will be processing data instead of blood. Initially, the group hopes to mimic individual hepatocyte cells in computer code. Small, self-contained sets of instructions will be used to represent the different parts of the cell, such as receptors, proteins and enzymes. Cells will communicate with each other and pass around information about the type of “blood” or data passing through them. As a result, groups of cells will be able to adapt to the data they find themselves dealing with. The long-term plan is to build an entire liver. Instructions about the overall “organ” will incorporate those governing the functions of the lobules, which are partly determined by what the smaller hepatocytes find in the data reaching them. Paton says the result should be an information processing system that is capable of dealing with the kinds of fuzzy data that today’s crop of computers struggle to deal with, such as image recognition or spotting trends in huge databases. It may also lead to more efficient operating systems for parallel processing machines. Hans Paul Schwefel a systems analyst at the University of Dortmund,