Twitter | Search | |
Lucas Moore
Holy crap. Huge bug uncovered in computational chemistry software because different operating systems sort files differently and the published scripts don’t handle it well. If you do or rely on calculated NMR chemical shifts, this is a must-read.
The bioactivity-guided examination of a Leptolyngbya sp. led to the isolation of leptazolines A–D (1–4), from the culture media, along with two degradation products (5 and 6). Density functional...
ACS Publications ACS Publications @ACSPublications
Reply Retweet Like More
Nguyen Doan Oct 8
Replying to @LucasCMoore
Well, while it is funny and interesting to see this glitch, I can't really see whether this result is significant. 1. Calculations are mostly done on supercomputing system, which uses Linux as operating system. So Linux maybe considered the standardized one here.
Reply Retweet Like
Nguyen Doan Oct 8
Replying to @LucasCMoore
3. Who on the earth relies to computational result 100%???? While many people did use computational chem to predict the NMR of some structures, they still need to synthesize those structures to verify it at the end. A good example is the synthesis of Dichrocephone of Christmann
Reply Retweet Like
Ash Jogalekar Oct 8
Replying to @LucasCMoore @Dereklowe
Nice paper, but would the difference between 172.7 and 173.2 happen to be within the error bars of the calculation?
Reply Retweet Like
Lucas Moore Oct 8
Replying to @curiouswavefn
That’s just one example, but the point is that the scripts can swap any two conformations’ energy levels – they might be much further apart than that.
Reply Retweet Like
Nicholas F Chilton Oct 8
Only read abstract/ TOC, but: is < 0.5% difference really "huge" in this context? Bug, sure, problem, sure, but changing conclusions? Really? What's the experimental line width of the resonance? thoughts?
Reply Retweet Like
Lucas Moore Oct 8
Imagine your analytical balance randomly swaps measured masses of samples. If they’re all close, you can get away with it. But they’re not always close. Predictive NMR isn’t perfect, and there are error bars. But those error bars shouldn’t be widened by scripts scrambling data.
Reply Retweet Like
🇨🇭 Heinz Holler 🛠 @toxomat@social.tchncs.de Oct 8
Programmer assumed result of Python "glob"-function is sorted, when in fact it is not. Correct results are coincidences and not a matter of OS but of filesystem, strictly speaking. Also, ACS violates copyright by publishing fixed code.😂
Reply Retweet Like
Karthik Kumar Viswanathan Oct 11
Programmer A: I'm using X. Why is this not sorted? Programmer B: It was never meant to be sorted, not at least how you wanted it to be. Programmer A: Bug! System is bad!
Reply Retweet Like
Ernesto Oct 8
Replying to @wokus_dei @LucasCMoore
I’m not, just a lowly software dev in a medtech-y team
Reply Retweet Like