“imagine a future in which people with more money thanks to her privileged genes longer and healthier life.”
The geneticist Jennifer Doudna and Samuel Sternberg, in “A Crack in Creation” (2017)
in these days of scientific reports, their cross-connections and implications are not immediately apparent. A current example is the recent issue of “Nature”.
it’s about a Protein called Interleukin-2. It promises to be – in theory – highly interesting new possibilities for immunotherapy against certain types of cancer. Interleukin-2 but has a major drawback: It is, although a natural substance of the body, in larger amounts very toxic.
more Stable, less harmful, completely different-looking
In the “Nature”article reports, Daniel-Adriano Silva, and about 20 ladies and gentlemen, as you are a Designer Version of Interleukin-2 have evolved, these unfavorable properties, apparently, are reduced or completely Disappear. In mice to intestinal and skin cancer treatment. An artificial, a body’s own substance similar to Protein, which could be used in a few years, possibly, in cancer therapy, created in the laboratory, and, now it comes: with a “komputationalen approach”.
The design of protein was produced in a complex iterative process alternately on the computer, and physically in the laboratory. In the end, it didn’t look like the template, but had a less harmful properties and was also more stable than the Original one.
The Software, with the help of the Protein was Calculated implemented, it is called Rosetta, and its inventor is one of the authors of the article. David Baker of the University of Washington is one of the pioneers of this fairly new discipline that we might call computational biotechnology: analysis, prediction, and design of protein structures with the help of complex Software. Or, to put it differently: Here, the components are decomposed living organisms on the computer, predicted, interpreted and constructed.
Once been a living crane folds
the Rosetta spacecraft in the past years, always in the context of the CASP competition used, was here a few weeks ago already once a topic: it is about which Team can say the best from given DNA sequences, three dimensional protein structures in advance. A task whose complexity has explained Nerd Cartoonist Randall Munroe, once said: “Did you fold a crane out of paper?” – “Yes” – “Then you can imagine right now, you find out need to which folds are necessary to a real live crane”.
At the recent CASP competition, there was a non-surprise winner. The ten-strong Team of Google – more precisely, the Alphabet, a subsidiary of deep mind suggested the cream of this discipline, by a large margin. Deep mind has developed a System called AlphaFold based on artificial neural networks. The technology, therefore, the give us the rapid progress in the area of so-called artificial intelligence.
crisis of meaning for protein moth
The victory of deep mind crashed many a Star of the international protein moth scene in a crisis of meaning, such as derGenetiker Mohammed Al-Quraishi in a read blog entry, the values reported. First of all, many of our colleagues had feared, deep mind got rid of him the people with a brilliant new approach. Then relief: The method and the tool are not new, however the basic approach. Then you’ve got started, the success of small persuasion, and the resources used.
This reminds me a bit of the reactions from professional Go players on the initial success of deep minds AlphaGo against the European Champions in the game: Well, it wasn’t so great now. All just computing power. Real pros are still better. What happened next is now well known.
KI can’t apply now
The deep mind Team is the only group of researchers, the methods of machine learning on such tasks. Undoubtedly, the Google subsidiary has in doubt more money and computing power as a University researcher. The bottom line from the result, however, is in my eyes a: AI can now such things. You forget-in speaker with voice recognition and self-driving cars: perhaps The most drastic changes is KI soon in the area of biotechnology.
DNA data, protein structure data. The problems to be solved are extremely complex, it is multi-causal, confusing relationships, structures and patterns. Lots of things so that the AI of today and can handle the morning well. Learning systems are not just built and then you are finished – if you do it more and more of them screwed in, you will also continue to be better.
A scientific Revolution, of which we are not to dare to dream
This, to biotechnology and machine Learning, will, in the coming years, probably a scientific Revolution, we dare to dream yet. The disciplines in which this Revolution will take place, are not even prepared yet. The Harvard researchers system biologist Al-Quraishi writes: “Competitive paid research engineers with Expertise in the area of Software and Informatics lack in academic laboratories almost completely, despite the Central role they play in industrial research laboratories.”
So it can happen that a small group of talented takes on outsiders, suddenly, the lead in a decades-old research tradition. Not only the academic scene was surprised, but apparently, the big pharmaceutical companies – also seem, even though they had the necessary money, to sleep in the matter of machine Learning as a research tool so far.
A machine that designs people?
The change probably just. Pfizer, Merck, Novartis, and co. will now contest a lot of highly paid Jobs for AI professionals. But the deep mind-people will not stop at this prestigious thing to continue to work.
companies and universities will start soon, or at least with learners of machines advanced design of molecules. Next, machines will be making proposals for gene therapy interventions. The patient, whose life may be saved or extended, will be the very right.
And at some point comes somewhere perhaps, from a secret laboratory on the planet, the first Designer DNA from the machine.