5 ms·

The headline is sensationalized. They did a virtual screen of a large compound library. QSAR / QSPR / Virtual Screening has been around since the late 1950s.

The headline is sensationalized. They did a virtual screen of a large compound library. QSAR / QSPR / Virtual Screening has been around since the late 1950s. The secret sauce here was the large experimental dataset - on the order of 10^5 compounds - they generated to train the model. Maybe the explainability part is kind of novel for a neural network based approach; but not clear that you couldn’t identify substructure classes better with a pure bioinformatic approach.

wolverine876 3y ago
The OP says something different: "The insight here was that we could see what was being learned by the models to make their predictions that certain molecules would make for good antibiotics," James Collins, professor of Medical Engineering and Science at the Massachusetts Institute of Technology (MIT) and one of the study’s authors, said in a statement. ... "What we set out to do in this study was to open the black box. These models consist of very large numbers of calculations that mimic neural connections, and no one really knows what's going on underneath the hood," said Felix Wong, a postdoc at MIT and Harvard and one of the study’s lead authors. > not clear that you couldn’t identify substructure classes better with a pure bioinformatic approach. Lots of things aren't clear, but what supports your particular claim? Has it ever been done? And is it addressed in the article?
- hedora 3y ago
  They’re saying that traditional computational drug discovery pipelines do all the stuff the authors are claiming. So, yes, it’s been done at scale for decades. The article makes all sorts of incorrect claims. For example, it says it has been 60 years since the last antibiotic was discovered, and that this works enables computational screening of antibiotics for the first time. Here’s work from 2022 that used conventional computational screening to discover a novel antibiotic: https://phys.org/news/2022-10-discovery-antibiotic-resistant-pathogens.html https://phys.org/news/2022-10-discovery-antibiotic-resistant...
  - wolverine876 3y ago
    These claims on HN, about every OP, really lack credibility for me. Either HN commenters are the only smart people in the world or we are misleading each other. At the same time, we are dissuading each other from engaging in a world of intellect, knowledge and innovation. HN is a forum of intellectual curiosity. The near-universal critiques are the opposite of that.
    - COGlory 3y ago
      I don't know what to tell you. New antibiotics are discovered pretty often. They aren't brought to market, but they certainly are discovered. This headline is just wrong.
    - Jensson 3y ago
      > These claims on HN, about every OP, really lack credibility for me. Do you trust journalists that much? Have you ever read an article about something you are an expert on? You shouldn't trust articles that much. I trust HN discussions much more since there are plenty of people who will weigh in when things are wrong or misrepresented, unlike these articles. Edit: Gell-Mann Amnesia effect, that was what this is called. Don't trust the media just because you aren't an expert.
      - wolverine876 3y ago
        You've lept to a conclusion about what I trust and then how that putative trust drives my reasoning. I'm not the imaginary person you are addressing. > I trust HN discussions much more since there are plenty of people who will weigh in when things are wrong or misrepresented, unlike these articles. Why are those people, with no responsibility, little expertise, etc., more trustworthy than other people? It seems like the common bias of social media - for some reason, people trust strangers with no credibility, which exposes them to the enormous amount of misinformation and just BS that now floods our world. If the article author or a professor wrote the same thing in a HN comment, it seems like you'd believe it.
        3y ago
Maxion 3y ago
I would recommend reading the actual paper, and not the press release. The main novelty here and what the paper is about is their method, not the discovery of the antibiotic.
- pfisherman 3y ago
  Chemprop is not new, it has been around since at least 2020.
la64710 3y ago
Exactly , it seems the overall process of sorting and filtering vast amount of data remains the same only that the AI component is an additional tool to the toolchest to accelerate the process.

wolverine876 3y ago

The OP says something different: "The insight here was that we could see what was being learned by the models to make their predictions that certain molecules would make for good antibiotics," James Collins, professor of Medical Engineering and Science at the Massachusetts Institute of Technology (MIT) and one of the study’s authors, said in a statement. ... "What we set out to do in this study was to open the black box. These models consist of very large numbers of calculations that mimic neural connections, and no one really knows what's going on underneath the hood," said Felix Wong, a postdoc at MIT and Harvard and one of the study’s lead authors. > not clear that you couldn’t identify substructure classes better with a pure bioinformatic approach. Lots of things aren't clear, but what supports your particular claim? Has it ever been done? And is it addressed in the article?

hedora 3y ago
They’re saying that traditional computational drug discovery pipelines do all the stuff the authors are claiming. So, yes, it’s been done at scale for decades. The article makes all sorts of incorrect claims. For example, it says it has been 60 years since the last antibiotic was discovered, and that this works enables computational screening of antibiotics for the first time. Here’s work from 2022 that used conventional computational screening to discover a novel antibiotic: https://phys.org/news/2022-10-discovery-antibiotic-resistant-pathogens.html https://phys.org/news/2022-10-discovery-antibiotic-resistant...
- wolverine876 3y ago
  These claims on HN, about every OP, really lack credibility for me. Either HN commenters are the only smart people in the world or we are misleading each other. At the same time, we are dissuading each other from engaging in a world of intellect, knowledge and innovation. HN is a forum of intellectual curiosity. The near-universal critiques are the opposite of that.
  - COGlory 3y ago
    I don't know what to tell you. New antibiotics are discovered pretty often. They aren't brought to market, but they certainly are discovered. This headline is just wrong.
  - Jensson 3y ago
    > These claims on HN, about every OP, really lack credibility for me. Do you trust journalists that much? Have you ever read an article about something you are an expert on? You shouldn't trust articles that much. I trust HN discussions much more since there are plenty of people who will weigh in when things are wrong or misrepresented, unlike these articles. Edit: Gell-Mann Amnesia effect, that was what this is called. Don't trust the media just because you aren't an expert.
    - wolverine876 3y ago
      You've lept to a conclusion about what I trust and then how that putative trust drives my reasoning. I'm not the imaginary person you are addressing. > I trust HN discussions much more since there are plenty of people who will weigh in when things are wrong or misrepresented, unlike these articles. Why are those people, with no responsibility, little expertise, etc., more trustworthy than other people? It seems like the common bias of social media - for some reason, people trust strangers with no credibility, which exposes them to the enormous amount of misinformation and just BS that now floods our world. If the article author or a professor wrote the same thing in a HN comment, it seems like you'd believe it.
      - 3y ago

Maxion 3y ago

I would recommend reading the actual paper, and not the press release. The main novelty here and what the paper is about is their method, not the discovery of the antibiotic.

pfisherman 3y ago
Chemprop is not new, it has been around since at least 2020.

la64710 3y ago

Exactly , it seems the overall process of sorting and filtering vast amount of data remains the same only that the AI component is an additional tool to the toolchest to accelerate the process.