@L_howes I wrote up a first blog post about my experiences. Moving from one session to another indeed works a lot better online. I've always been a bit more of a CINF person than COMP, and also this time, did not find the COMP sessions very attractive. Well, that's not entirely true, it *is* attractive and interesting, but it just doesn't excite me enough
Online participation does make it a lot easier to just probe a remote session. Nice to join some #chemistry history talks at #ACSFall2023 :)
@L_howes okay. Since posts spread a bit more slowly here on #mastodon (slow is good), and not everything is visible on every server (community standards are essential), I wasn't sure I was receiving all the online #ACSFall2023 vibes.
But now we know at least that #chemistry has not landed on #mastodon (yet). Better luck next time :)
@egonw I think you were almost singlehandedly posting #ACSFall2023 here but I personally really appreciated it. Especially to get a sense of sessions I wasn't in. And virtual probably helped you follow COMP and CINF without shuttling across the city and back all the time
@L_howes not so many people here on Mastodon who adopted conference posting (*), but did you enjoy the #ACSFall2023 posting here? I did my best to keep up with all the fun stuff
Although I was unable to attend this Fall, I am so happy that 4 students from my group were able to share their awesome science at #ACSFall2023 🧪😎
Never misses a chance to talk about Feynman #acsfall2023
I mean, I even here now "we like to use version control because..."
It is so cool that things that people from the @blueobelisk movement have been developing, promoting, are pretty normal at an ACS meeting now! #ACSFall2023
#ACSFall2023 was a blast, and it has been great to show how multimodal
foundation models and hybrid cloud computing will support scientists in the lab of the future.
We ran 47 experiments with our fantastic visitors, resulting in 1512 model predictions and 2 retraining rounds.
A lot of thrashed cold brew coffee, but it was definitely worth it.
Kudos to the whole @ibmresearch team.
Stay tuned for exciting developments, and check out the Lab that Learns project: https://lnkd.in/enCNMYKW.
I have no idea if any of the speakers is here on #mastodon yet, but thank you, thank you! I enjoyed all the presentations I watched. And the amount of #openscience was mind blowing! preprints, CC-BY work, GitHub repositories, it was all there. Well done, #chemistry! #ACSFall2023
but I have to say, it has been a awesome again. But this years has been double awesome, my first American Cheminformatics Society meeting! #ACSFall2023
I am also a bit sad that the presentations (and I'm sure I was just like that myself) do not really indicate what they do better than people before. There is a high level of "we did it this way". Sure, but I want to hear in your talk why this is better, instead of hoping the articles tell me that. #ACSFall2023
I know I was the same, it's hard to know the past 30 years of research, but in the "AI for organic synthesis" session, I am sad to see how few people seem to know the work from Aires-de-Sousa :( #ACSFall2023
quick update on that in four #chemistry data in #wikidata plots, melting point, boiling point, dipole, and IP (looking that numbers, just me right now)
The 32k melting points are from NIOSH and particular Jean-Claude Bradley #ACSFall2023
(see my Sunday talk, basically)
while not that different, the scale is impressive. We're not talking about databases of 10-10,000 compounds, but of databases of 10,000,000-10,000,000,00 chemical compounds.
But I also observe the all work with SMILES and computed properties. experimental data is, I guess, still locked up?
/me opens Wikidata again to add some more experimental data... it has to start somewhere, doesn't it?
both #ACSFall2023 session are now talking about the #cheminformatics of organic #chemistry reactions
both sessions are basically classical cheminformatics but with a lot more data. and what we saw for chemometrics too, we see that people have started thinking about ensembles and not molecules anymore. not at the model level. and like in chemometrics and in biology, we apply a difficult to grasp model to individual entities and see if it makes sense. N=1,2,3 evaluation of a N^x complex space. Latent spaces won't help. #chemistry #ACSFall2023
last #ACSFall2023 day. Going to try to follow to related sessions again, "AI in organic chemistry" (where more and more open data is becoming available) and "deep learning / chemical space", but that one seems stalled right now?
final one I can watch was KT-474, a racemic mixture: https://www.wikidata.org/wiki/Q121502359
Something really funny with the CIP annotation happening here around the hexane ring... I only change the stereocenter of one atom and kept that of the hexane ring unaltered, but R,R becomes R,S...
