In May last year, a supercomputer in San Jose, California, read 100,000 research papers in 2 hours. It found completely new biology hidden in the data. Called KnIT, the computer is one of a handful of systems pushing back the frontiers of knowledge without human help.
KnIT didn't read the papers like a scientist – that would have taken a lifetime. Instead, it scanned for information on a protein called p53, and a class of enzymes that can interact with it, called kinases. Also known as "the guardian of the genome", p53 suppresses tumors in humans. KnIT trawled the literature searching for links that imply undiscovered p53 kinases, which could provide routes to new cancer drugs.
Having analyzed papers up until 2003, KnIT identified seven of the nine kinases discovered over the subsequent 10 years. More importantly, it also found what appeared to be two p53 kinases unknown to science. Initial lab tests confirmed the findings, although the team wants to repeat the experiment to be sure.
(Score: 0) by Anonymous Coward on Sunday August 31 2014, @03:19AM
42
(Score: 0) by Anonymous Coward on Sunday August 31 2014, @11:29AM
This is no normal super computer. Instead of using init it uses initiative.
(Score: 0) by Anonymous Coward on Sunday August 31 2014, @03:48AM
This sounds much like a search engine being able to put together a bunch of different sources based on a single search term to make desired information more convenient to the searcher. That search engines help people find and piece together relevant information so that they can more quickly and easily do things that wouldn't have been as easy before is nothing new.
(Score: 2) by SlimmPickens on Sunday August 31 2014, @03:56AM
In a way you're right, but I don't think there's any combination of Google operators that can come up with
what appeared to be two p53 kinases unknown to science
.
It's got to filter a hell of a lot of false leads to do that.
(Score: 2, Funny) by Horse With Stripes on Sunday August 31 2014, @11:33AM
Have ever tried using Google's "kinases: unknown" search operator? ;-)
(Score: 4, Informative) by tynin on Sunday August 31 2014, @04:28AM
It is true, but the new level of awesomeness in this is at least two part....
1. Thanks to machine learning, we can re-eval everything we have gathered significant data on. Where we feel we understand a myriad of properties (which might be greater than a single mind can ingest), we can uncover what should have been obvious (or not) but got lost in the mix, or...
2. the right person who thinks to ask the right questions (writes the right algorithm) might just come away with a new understanding that could propel us all forward, to being able to keep this (biological/metaphysical) game in play.
While we aren't there today... perhaps soon. As they say, We are standing on the shoulders of giants.
(Score: 3, Interesting) by Anonymous Coward on Sunday August 31 2014, @05:29AM
> While we aren't there today... perhaps soon. As they say, We are standing on the shoulders of giants.
If they are able to make this technique work for anything even remotely close to a generalizable form then it will turn those millions of essentially unread scientific papers that the publish-or-perish system creates into a fertile resource. It may also make cases of negative results more valuable - all those experiments from drug companies that failed to pan out and normally get binned instead of published might reveal something else. But I think your analogy is reversed - it will be more like giants standing on the shoulders of millions of ants.
(Score: 2) by Tork on Sunday August 31 2014, @06:38AM
🏳️🌈 Proud Ally 🏳️🌈
(Score: 4, Interesting) by kaszz on Sunday August 31 2014, @03:56AM
Writing a parallel science paper coded for (semantic?) computer analyze is probably the way of the future. There's just too much information to deal with for humans. First we use computers to get information. Now that they are powerful enough we use the to analyze the same. The next level of information use.
But don't count biochemists as out of work just yet. They just get more productive which causes the need for even more biochemists because they can do even more wonders.
As for cancer suppression. I hope there will be a more clever way to cure it that the current method which is quite medieval and originally based on mustard gas Cl--S--Cl (mustine Cl--N--Cl) poisoning being worse for cancer cells than for good cells.
(Score: 5, Insightful) by mcgrew on Sunday August 31 2014, @01:14PM
First we use computers to get information. Now that they are powerful enough we use the to analyze the same.
Incorrect, we've been using computers to analyze information since computers were invented in the 1940s. I used computers to analyze data for decades before I retired. Using computers to obtain data is relatively new; in the beginning, all data had to be entered by hand.
But don't count biochemists as out of work just yet.
Of course that's correct; data analysis isn't thought. A computer doesn't know anything any more than a printed book does.
As for cancer suppression. I hope there will be a more clever way to cure it that the current method which is quite medieval and originally based on mustard gas Cl--S--Cl (mustine Cl--N--Cl) poisoning being worse for cancer cells than for good cells.
Correct, but misleading. They have far better chemicals than that now, [wikipedia.org] and far more treatments. Many cancers are no longer automatic death sentences.
And if you think medicine is primitive now, you should have been alive fifty years ago. Fifty years from now, today will be primitive,
mcgrewbooks.com mcgrew.info nooze.org
(Score: 2) by kaszz on Sunday August 31 2014, @01:33PM
Incorrect, we've been using computers to analyze information since computers were invented in the 1940s.
Perhaps we should call it sorting etc. Those computers weren't powerful enough for any deep analysis. Now the computing capability is of another magnitude which opens the door for completely new methods.
There are better Chemotherapy. But it's still based on the same origin. And it poison the whole body rather than targeting the cancer cells and leave the rest intact.
Yeah, 50-years ago was primitive. The questions is always if the remedies will show up before you need them.
(Score: 0) by Anonymous Coward on Monday September 01 2014, @01:52AM
Too much information isn't the problem. Essays are the problem. Most of those essays could be condensed into a 5lines wiki topic edit and a few side-notes (preferably hyper-linked) to relevant lab results with detailed methods and such.
(Score: 0) by Anonymous Coward on Sunday August 31 2014, @04:13AM
"Initial lab tests confirmed the findings"
No, initial tests are consistent. They need to be repeated in order to confirm that the kinases directly phosphorylate p53.
(Score: 2, Redundant) by TGV on Sunday August 31 2014, @05:28AM
If a group on two hundred students had been reading through all these papers, would the headline have been: Students Make Discoveries that Scientists Can't? If it had been two hundred scientists, would it have been: Scientists Make Discoveries that Scientists Can't? A better headline would have been: body of literature too large, and that's the case since a long time. In larger fields, many studies have already been done, but nobody knows.
(Score: 1, Interesting) by Anonymous Coward on Sunday August 31 2014, @05:35AM
Just what precisely is your complaint here?
The computer was able to synthesize something new from a set of information that is too big to fit into any single human brain. Synthesis is a key characteristic of intelligence. However narrow the application in this case, it is essentially an aspect of AI research.
(Score: 5, Insightful) by janrinok on Sunday August 31 2014, @07:04AM
If you read the provided link, you will find that the title was provided by the original source material. I do not think that it sensationalizes supercomputer science and it accurately describes what took place.
I accept that when the editors get something wrong they should be called out for it - and we often are. If we make a mistake then we try to both correct it and apologise for the error. But there is a growing tendency for nit-picking of minor detail in stories. If we are getting that much wrong then perhaps other members of the community should step up to the plate and shoulder some of the burden. But take note - nobody on SN gets paid anything, we all do it because we believe in this site, we are all trying to do our best, and I have had 2 weekends off since May.
We would rather see a comment on the story itself than read that someone who has spent a lot more time than we might have available has thought of a different wording to the title.
Just my 2 cents worth...
(Score: 1) by TGV on Sunday August 31 2014, @07:33AM
> the title was provided by the original source material
I know e.g. reddit insists on copying the title of an article in a link, but there's absolutely no need for that. Here are two other descriptions of the same research with rather different titles:
https://www.bcm.edu/news/research/automated-reasoning-hypothesis-generation [bcm.edu]
http://www.hpcwire.com/2014/08/28/watson-based-tool-automates-discovery/ [hpcwire.com]
> We would rather see a comment on the story itself than read that someone who has spent a lot more time than we might have available has thought of a different wording to the title.
Passive-aggressive insults from editors to readers. Nice. (BTW, I know this is a one too; it's just to make you experience what such a phrase can provoke).
I disagree (mildly) with your statement anyway. Editors take it upon themselves to spend time on reading the article and making an entry. They're the reason people visit soylentnews. We visit sites like this because we can't or don't want to spend much time on reading through all the announcements. Soylentnews had the vague promise of being more down-to-earth and less "repeat the PR in order to get clicks" than other sites.
(Score: 3, Informative) by Anonymous Coward on Sunday August 31 2014, @08:13AM
> Editors take it upon themselves to spend time on reading the article and making an entry. They're the reason people visit soylentnews.
Perhaps that's why you come here, but the reason Soylent was created was for the community as expressed in the comments section because slashdot was making changes that minimized their community.
If you want news summarized to standards that you approve of you should considering hiring someone to do that for you.
Or more succinctly, quit being an ass.
(Score: 3, Interesting) by TheRaven on Sunday August 31 2014, @09:27AM
sudo mod me up
(Score: 3, Insightful) by janrinok on Sunday August 31 2014, @11:33AM
My comment wasn't intended to be an insult of any kind - and apologies if it came across as one. As I mentioned in my previous post, yours is not the only comment criticizing the title, subject matter, or layout of the material that we post. We know that we haven't got it right all the time yet and we do take note of corrections and suggestions - but we get far more encouragement when we see comments on the story rather than those pointing out different ways it could be titled or whatever. There are many different ways it could be written, but sometimes I go for one which (in my opinion) summarizes the summary succinctly and accurately.
My personal view is that the two alternatives that you suggested are both acceptable but not significantly better than the title I used. You have the right to disagree with my opinion.
I may have volunteered to be an editor, but when it stops being a pleasure and becomes a chore, and especially a chore for which we seem to be criticized for trivial reasons (and I do think that the title, in this instance, is a trivial point), it is hard to keep being motivated and not to consider simply leaving it to someone else. Except there are very few others who seem to want to help.
None of this is intended as an insult to you or any other members of our community - so please do not take it as one.
(Score: 2) by Horse With Stripes on Sunday August 31 2014, @01:17PM
By all means, please submit stories that you've summarized to meet your personal standards and assigned headlines that you feel accurately reflect the content of the stories you link to.
The easiest way to improve the level of content in this community is to lead by example.
(Score: 0) by Anonymous Coward on Sunday August 31 2014, @02:38PM
Scientists are nit-picky. Don't take it as a bad thing. Scientists constantly try to catch themselves and others in their wording to prevent overstatements or be more precise: "significant difference" - did they do statistics?; "p53 kinase" - direct or indirect?; "activates the kinase" - did they do a kinase assay or just look at downstream targets?; "drives transcription" - does it directly bind the promoter/enhancer in cells or did they just correlate transcriptional changes?
(Score: 2) by janrinok on Sunday August 31 2014, @06:13PM
er, I don't know. Is 42 a good answer?
You make a valid point - but the questions that you ask are way beyond any area of my expertise! Sorry that my summary must seem to you as though it was written by a child./p.
(Score: 2) by SlimmPickens on Monday September 01 2014, @12:26AM
er, I don't know. Is 42 a good answer?
LOL! Oh for a mod point.
(Score: 2) by AnonTechie on Sunday August 31 2014, @07:50AM
System to Automate Reasoning, Hypothesis Generation:
http://soylentnews.org/article.pl?sid=14/08/26/2351257 [soylentnews.org]
Albert Einstein - "Only two things are infinite, the universe and human stupidity, and I'm not sure about the former."
(Score: 0) by Anonymous Coward on Sunday August 31 2014, @08:19AM
You are absolutely right.
But that summary was so dry that my eyes glazed over halfway through. I'm sure it is meaningful to an expert in the field, but we can't all be experts in the fields of all posted stories. A little story-telling goes a long way.
(Score: 0) by Anonymous Coward on Sunday August 31 2014, @09:25AM
Nice new powers then. Unfortunately I don't expect them to be used for good. Fortunately access to science is limited. On the other hand this will only make it more likely that who ever spends the big bucks to dig out publications will use any insight for private profit rather than common good...
Can we do this using distributed computing? That way we can all be Dr. Evils.
(Score: 2) by opinionated_science on Sunday August 31 2014, @04:38PM
Isnt't this just a work generation machine?
My point being, that publications are an specific form of communication, but still represent human work of some order.
One of the reason this "new biology" is hidden is the way in which scientific research is carried out, biology in particular. Grants are written for specific targets, and investigators (senior on down) channel resources to solve that target.
Cancer is complex. And Google and the like are critically important. But all this shows is that a human can read everything and needs a machine to propose likely connections.
Especially when you talk about "systems biology" we are only just starting to get a handle on it...