计算机专业时文选读之三十二

Computation Comes to Life

For years biologists have used computer models and high-performance computers to simulate and understand living processes. More recently, computer scientists have drawn inspiration from biology to immunize information systems against malware and to create algorithms that mutate without human intervention. In all such cases, the underlying computer architecture has remained traditional and unremarkable——software running on silicon-based digital processors.

But now researchers are taking the marriage of computer science and biology to a remarkable new level, turning cells into living computers with programmable DNA and biochemical memories, sensors, actuators and intercellular communication mechanisms.

Chip-making processes today place atoms of silicon and dopants——impurities added to define the chip's electrical properties——crudely but well enough to make the chips work. As circuits shrink, however, it's getting harder to put the atoms, particularly the dopant atoms, in exactly the right places.

But biological processes for millions of years have been able to place single molecules and atoms in precisely the right order and locations.

Rather than wait centuries for conventional engineering to catch up, Thomas Knight, an MIT researcher and a pioneer in the field and researchers at a handful of universities want to ride on the back of biology or, more precisely, inside the cell. Knight and a group of graduate students are building a tool kit of what they call BioBricks, standard parts that can be used to build programmable organisms.

Each of some 400 BioBricks is housed in a little vial of liquid containing copies of a carefully chosen and well-understood section of DNA. Each DNA fragment can mimic in some way the operations of conventional computer circuits. BioBricks can be used individually to perform very simple tasks, or they can be spliced together to do higher-level work. They allow someone to build programmable organisms without understanding the underlying biology.

There are BioBricks that act as logic gates, performing simple Boolean operations such as AND, NOT, NOT AND, OR, NOT OR and so on. For example, the AND BioBrick generates an output signal when it gets a biochemical signal from both its inputs, whereas an OR BioBrick produces a signal if it gets a signal from either input.

These biological components work extremely slowly by the standards of conventional computers, performing their functions in seconds or minutes rather than nanoseconds, and Knight says they are unlikely ever to exceed millisecond-level performance. But that doesn't mean you couldn't use biological components to produce, say, carbon nanotubes, that in turn could be used to build molecular-scale high-performance computers.

Or, Knight says, it's possible that living factories made from BioBricks could help build ultradense silicon chips by replacing the troublesome dopant atoms at just the right points on a silicon lattice.

Ron Weiss, a former student of Knight's and now a professor of electrical engineering and molecular biology at Princeton University, is working on digital logic inside cells and intercellular communications. He says it will be a long time before synthetic biology contributes directly to computer science. “But eventually we might come up with an abstraction that allows you to program billions of little biological computing elements that are not robust at all and don't have a lot of resources,”Weiss says, “and that might be a useful paradigm for programming certain kinds of silicon-based computational devices.”

Scientists at the University of Alberta in Edmonton are trying to develop a plant whose leaf shape or flower color changes when a land mine is buried below it. Roots would have to be genetically altered to detect explosives traces in the soil and to communicate that information to the leaves or flowers.

That will require some kind of sensor circuits in the plants' root cells, plus an actuator circuit in the leaf or flower cells, with little real computation in between. But, Knight says, one can imagine more-sophisticated computational engines inside a plant's cell that would, for example, cause the plant to bloom on Mother's Day or prepare itself for frost or drought based on warnings input by human weather forecasters.

But he's clearly uncomfortable speculating about miraculous applications of synthetic biology. A great deal of effort must first go into developing the kinds of design and measurement tools and methods that conventional engineers take for granted.

The ability of biological circuits to self-replicate makes synthetic biology unique among all engineering disciplines, Knight says. “Tremendous power comes from that, and some dangers,” he says.

Researchers at MIT are limiting their work to two kinds of agents. The first are natural agents that are 100% safe, and the second are engineered organisms “not known to consistently cause disease in healthy adult humans,”the government's definition of Biosafety Level 1 on its four-level scale of infection dangers. And, Knight adds, his work involves simplifying organisms, not adding features that could make them dangerous.

The greater danger in synthetic biology, Knight says, comes from the possibility that others will exploit it for evil purposes. “All powerful technologies are dangerous, and we are creating a powerful technology,”he says. “Our best defense is our ability to do it faster, better and cheaper than anyone else.”

计算进入生命科学

多年来，生物学家利用计算机模型和高性能计算机模拟和了解生命过程。最近，计算机科学家从生物学获得灵感，(开发出)使信息系统对恶意软件免疫(的技术)，还编制出无需人工干预就能变异的算法。在这些情况下，基础的计算机架构仍是传统的和常态的——软件运行在基于硅的数字处理器上。

但现在，研究人员在一个全新的层面上将计算机科学与生物学联姻，把细胞变成拥有可编程DNA与生化存储器、传感器、激励器和细胞间通信机制的有生命的计算机。

今天的芯片制造工艺是将硅和掺杂物(加入的杂质决定芯片的电气特性)的原子进行排列，原始但足以使芯片工作。然而，随着电路缩小，将原子、尤其是掺杂物的原子放到正确的地方越来越难。

但是，经过几百万年(进化的)生物过程能以非常精确的序列和位置放置单个分子和原子。

此领域的先驱、MIT的研究人员Thomas Knight，以及一些大学的研究者不想再等上几个世纪让常规的工程技术赶上来，他们要驾驭生物学，更准确地讲，进入细胞内。Knight和一群研究生正在开发称之为BioBricks的工具套件，它们是能用来制造可编程有机体的标准部件。

他们将大约400个BioBricks装在一只装有液体的小瓶里，里面包含着一个经仔细选择并完全了解的DNA片段的复制品。每个DNA片段能以某种方式模拟常规计算机电路的操作。每个BioBricks能用于执行非常简单的任务，或者能合在一起完成更高级的工作。它们也能让不懂基础生物学的人生成可编程的有机体。

有的BioBrick能起逻辑门的作用，能完成简单的布尔运算，如与、非、与非、或、 或非等等。例如，BioBrick中有两个输入都收到生化信号时就产生“与”信号输出，而BioBrick在其中任何一个输入上得到信号就产生“或”信号输出。

按常规计算机的标准，这些生物部件工作得非常慢，以秒或者分而不是以纳秒完成其功能，Knight认为，它们不可能超过毫秒级的性能。但这并不意味你不能用生化部件生成碳纳管，随后再用碳纳管制造分子级的高性能计算机。

或者，如Knight所说，用BioBrick构成生命工厂是有可能的，它通过替代硅晶格适当位置上令人讨厌的掺杂原子帮助制造集成度极高的硅芯片。

Ron Weiss是Knight过去的学生，现在是普林斯顿大学电气工程系的教授，他正在研究细胞内的数字逻辑和细胞间的通信。他认为，合成生物学能对计算机科学直接做出贡献还需要很长的时间。他说：“最终我们可能会形成这样一个概念，允许对数以亿计的小小的生物计算单元进行编程，虽然这些单元从整体上讲不是强健的、也不具有很多资源，但对某些基于硅的计算部件而言这可能是一种有用的编程范例。”

(加拿大)爱德蒙顿市的阿尔伯达大学的科学<