• Question: How does the robot force the production of unnatural proteins and I would like to know what is meant by making the robot behave itself cause that worries me

    Asked by AlexH to Kim on 11 Jun 2020.
    • Photo: Kim Liu

      Kim Liu answered on 11 Jun 2020:


      Hi again AlexH, thanks for this question ๐Ÿ™‚ It’s nice to be able to take more time on it! Leave comments at the bottom to continue the discussion ~
      ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
      I’ll first clarify how we produce unnatural proteins. Proteins are composed of building blocks known as amino acids – most organisms use 20 different naturally occurring amino acids to produce all of their proteins. These amino acids all have different chemical properties, and when you arrange them in a useful order, you get a useful shape and biological function. The neat thing about amino acids is that they chain up in exactly the same way, almost exactly like lego – the studs fit universally into the hollow of the next brick along. Approximately, a protein is like a stack of different lego bricks connected together, and they are assembled by a cool machine called the ribosome.
      ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
      The instructions for proteins are carried by the genetic code. DNA has 4 letters – A, C, G, T. The order of these letters tells the ribosome how to connect the amino acids together; it works like a language. For example, the sequence TCG usually encodes an amino acid called Serine. Our method of producing unnatural proteins involves giving the cells artificial cellular components which make the cells read TCG as an artificial amino acid – in this way, we produce unnatural amino acids by rewriting the genetic code. We may, for example, make the cell produce the same protein except that it has a fluorescent dye attached at the TCG position, so we could follow this protein around a cell. This has been done mostly in bacteria, but has been done in mice as well, which is pretty cool.
      ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
      Where the robot comes in: I have not explained anything of the ‘artificial cellular components’ that are required to make this work (I can if you would like; comment below!). What the robot is especially good at is helping us produce these cellular components through artificial evolution. We ‘force’ bacteria to make these artificial components by forcing them to save themselves from antibiotics that we apply. If they are able to make the component we want, they get to survive. It’s like making a random group of bacteria take a test, and only the ones that pass are allowed to survive – this is artificial evolution.
      ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
      I do find this quite hard to explain genetic recoding clearly and concisely haha – let me know how well I’m doing! There are a lot of steps involved, and it’s hard for me to know which steps to skip/simplify in the description.
      ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
      About the robot ‘behaving’ itself ๐Ÿ™‚ this is a great and thoughtful question; I use it as a euphemism for when the robot does ‘stupid’ things, so there’s nothing to worry about ~ But, you’ve raised an interesting thing about using robots in general – they do exactly what you tell them to do, and never apply ‘common sense’. So, one of the challenges of telling them what to do is carefully work out which instructions come under ‘human common sense’. Commonly, my robot tends to crash into things it doesn’t know are there, or hold things too tightly so it can’t let go of them. I know exactly why it does it, because these mistakes are technically ‘my fault’. It’s an interesting issue to deal with in computing and robotics – simulating ‘common sense’ is unbelievably difficult for computers; it requires very high levels of artificial intelligence my robot definitely does not have!
      ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
      This gets to a fairly challenging issue in artificial intelligence and machine learning – if a machine teaches itself how to do something, the human watching often can’t work out exactly what instructions the machine has written for itself. This is when people get worried, if the robot then starts to misbehave ๐Ÿ™‚
      ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
      Thanks for indulging my interest in this stuff ๐Ÿ™‚ If you are keen to, drop me a comment to continue the conversation ~

Comments