Hack-a-Mol | Hacker News Reader

Hack-a-Mol

chemapps.stolaf.edu

94 pointsbreck15 comments

Funny.

I have recommended this website (https://www.stolaf.edu/depts/chemistry/courses/toolkits/247/...) to many of my students to help them practice naming alkyl substituted alkanes.

The web-page has an "analyze" button that does a great job showing students how to find the longest carbon chain, which is then used to help determine the molecule's IUPAC name.

I've always wondered who was behind the site that I had recommended, but I was never curious enough to poke around and answer my own question.

When I saw "hack-a-mol" and stolaf.edu, I thought, "okay, must be the same person."

People like this (his name is Bob Hanson: https://www.stolaf.edu/people/hansonr/) are hidden gems. He's probably not super-famous, but he's made many great tools and I can only imagine that his students have benefited tremendously from his passion and skills.

Reminds me of Paul Falstad and his excellent circuit simulator tool. He originally wrote it as a Java applet, then gave permission to an Iian Sharp to have it rewritten in HTML5: https://www.falstad.com/circuit/

This is a very impressive site for teaching molecular file formats indeed! If there is one caveat that an old crystallographer might add...

  press ENTER, the 2D and 3D structures will update.

When going from the 2d connectivity graph to 3d there arise essential emergent properties, some of which can be specified, like stereochemistry (left or right chirality), and some which can be estimated with a lot of theory and additional calculation like multiple conformers (e.g. trans and cis), and thermal preferences (staggered vs eclipsed along rotable bonds), and onwards to intensive quantum chemistry for electron distribution which is so critical for design for enzyme binding. That is, "3d" at this level is very much an approximate single still image of a very dynamic system.

What a neat and timely teaching tool! Literally just an hour ago today I was reading an article in Chemical Review titled "Concepts of Artificial Intelligence for Computer-Assisted Drug Discovery" [0]. There was a lot of pretty standard stuff in there related to AI work but the real puzzle for me was trying to come up with a clean way to store and work with these data sets. I had come up with a few possible solutions that seemed decent but I knew there had to be a more standardized format. After struggling to get the correct search terms I decided to take a coffee break and then lo and behold, I come across this post. Thank you Mr. Hansom, you have successfully kept me busy the rest of the night :)

[0]https://pubs.acs.org/doi/10.1021/acs.chemrev.8b00728

That's great!

How do you represent a chemical structure mathematically? Is it a graph? Or something else? It seems more complicated than just a simple graph structure.

Mathematically: https://en.wikipedia.org/wiki/Schr%C3%B6dinger_equation

Is it a graph? Mostly yes - just ignore anything that's not covalent. It depends a bit on how complicated an interaction you're willing to allow an edge to represent, and how much of reality you're willing to ignore. A few examples of complications are linked below. The depictions may seem trivial, but for large molecules things quickly become ridiculous.

https://en.wikipedia.org/wiki/Resonance_(chemistry)

https://en.wikipedia.org/wiki/Pi_interaction

https://en.wikipedia.org/wiki/Antibonding_molecular_orbital

Anyone knows if this program is open source?

Molview is open source (and a great project imo) http://molview.org/

There is Openbabel, which was open-sourced by the folks from Openeye in New Mexico. Anthony Nichols is quite a character, and David Weininger was even more so.

The CACTVS software used on that website is related to Gasteiger at Erlangen.

I will second Openbabel as a reliable tool in my comp-chem pipeline.

What is going on? This was actually submitted 11 hours ago, but now re-appears under the same HN URL on the front-page, saying it was submitted only 1 hour ago?

On top of that all comments on it seem to have shifted by like 10 hours as well (busyant's comment is actually half a day old by now, not 40 minutes).

You can see the timestamps don't match up here: https://news.ycombinator.com/submitted?id=breck (says 11 hours ago).

Or here: https://news.ycombinator.com/threads?id=busyant (says 10 hours ago on his comment on this submission).

Likely the 2nd chance queue.

This is a process HN has devised to give selected submissions another chance at the front page. See:

https://news.ycombinator.com/item?id=11662380

It's kind of problematic that they roll back the time on user comments as well. This kind of "rewriting history" can lead to all sorts of issues.

How, specifically?

You don't see how lying about when something was posted can lead to problems? Just making sure I'm answering the right question.

The spectrum begins at me being slightly annoyed because I can't tell posts that were posted just now from posts that were posted hours ago.

It probably ends at people arguing in court over such things as prior art in patent cases, or someone trying to submit an online comment as evidence. Imagine Facebook mucking with timestamps... Often when is pretty important.

There's a lot of things in between there which I'll leave to your imagination.

In any case, changing meta information silently can be as bad as silently changing the content of posts themselves and will hurt your credibility as a platform. The reddit CEO did it as a joke once, and that already caused a massive shitstorm[1].

[1]: https://www.reddit.com/r/announcements/comments/5frg1n/tifu_...

Reply on news.ycombinator.com