On Dec 3, 2021, at 02:33, Suliman Sharif <sharifsulim...@gmail.com> wrote: > our generation gets to have a little fun where machine representation and > manipulation of chemical objects has been mostly figured out,
When was the current state of machine representation figured out? As a reminder, there's an infinite number of structures that SMILES can't handle. (Endohedral fullerenes and catenanes, to name two.) And 3D structures are at best a snapshot of a flexible structure. > We need a designated cheminformatic curriculum to explain what > cheminformatics is in the first place Does cheminformatics include its roots in library science? Or are those now different fields? (Given your use of "cheminformatic", I'm going to guess it will be some time before you get back to me on the puzzles. ;) > For example a first chapter could be teaching SMILES The thing about pedagogical selection of what to teach isn't "people should learn X" but rather "people don't need to know Y, and should learn X instead." What topics should receive less coverage, in order to teach SMILES, graph theory, etc? FWIW, while I lack the time and money for it, I've wanted to develop a "foundations of cheminformatics" book for a decade. The first chapter wouldn't be SMILES, it would be the Hill system, then chemical formulas, and only then SMILES. For example, http://dalkescientific.com/writings/NBN/parsing_by_hand.html or my PyCon video at https://archive.org/details/pyvideo_265___ply-and-pyparsing-93 . > So maybe implement your own InChl Key algorithm. Meaning, a grounding in graph canonicalization and the nauty algorithm? That's pretty esoteric even in cheminformatics. At least, I've not done that. > Another chapter could be how these titan databases (ZincDB, EnamineDB, > pubmed) were constructed and coded in doing rapid download etc. You learn a > lot doing this and I think it would be fun. Seems like the history of CAS (which is quite interesting, esp. if you include the politics) and knowing how to use SciFinder effectively would be more important to a lot of people. To be clear, CAS is very much NOT Open Data, and rather the antithesis of Blue Obelisk. But if your textbook is to have any inroads, you'll need to convince people whose future careers will likely be closely tied to ChemDraw and SciFinder that they need to learn other skills instead. As an aside to the list, in a hypothetical textbook, I would think 2D-based input is an important component. What's the FOSS equivalent to ChemDraw? ChemDoodle Web Component? JSME is not distributed in "the preferred form of the work for making modifications to it" (quoting the GPL). > What do you guys think? Any of these ideas hitting? I'm too jaded, in that I feel like I've seen all these ideas before and participated in some of them, and they never made an impact outside the core community. Which is fine! That's all they meant to do. But you want to 'refine education', which is far broader, and that's not an effort I feel like doing. Andrew da...@dalkescientific.com _______________________________________________ Blueobelisk-discuss mailing list Blueobelisk-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/blueobelisk-discuss