12
$\begingroup$

I have been testing a machine learning approach for molecular energy prediction. The current dataset that I have is QM9, which is consist of molecules with up to 9 heavy atoms.

I was wondering if anyone knows of available datasets that contain molecules with large numbers of heavy atoms. I will be testing ZINC, which has up to 38 atoms. Anyone knows of a larger dataset available?

$\endgroup$
  • $\begingroup$ In retrospect -- after giving providing an answer (molecules encoded as SMILES), while «user1271772» provided an other answer about molecules keeping geometries, does your question for a larger data set refer to a dataset containing molecules larger than 38 atoms, or a dataset with more molecules than ZINC? $\endgroup$ – Buttonwood Sep 1 at 0:18
  • $\begingroup$ Larger than 38 atoms. I was gonna say, but I appreciated your detail answer very much $\endgroup$ – Blade Sep 1 at 0:29
  • 1
    $\begingroup$ Considering your additional input, I added a second part to my initial answer. It is about an example which retains molecular geometries (similar to .xyz files) rather than a simplified representation as SMILES / InChi string. $\endgroup$ – Buttonwood Sep 1 at 21:09
9
$\begingroup$

The ISOL24 database (http://www.thch.uni-bonn.de/tc.old/downloads/GMTKN/GMTKN55/ISOL24.html) contains molecules with up to 81 atoms!

The other answer says that there's a database called "OE" with molecules that have up to 174 atoms, but it is "not yet publicly available".

$\endgroup$
  • $\begingroup$ Given the initial check in the work by Stuke et al. shown below, do you know about a similar survey about the entries in the ISOL24 database? $\endgroup$ – Buttonwood Sep 1 at 21:12
  • $\begingroup$ @Buttonwood: I do not understand your question. $\endgroup$ – user1271772 Sep 1 at 23:34
  • $\begingroup$ The question was if there equally is a plot like figure 2 extracted from doi/10.1063/1.5086105 which a) depicts which elements are present or/and b) the the distribution of number of atoms per molecule for the ISOL24 database. From its associated publiation in PCCP (Phys. Chem. Chem. Phys., 2017, 19, 32184, DOI: 10.1039/c7cp04913g) I retain it was setup to benchmark functionals and it were nice to compare it this way with the three sets below, too. $\endgroup$ – Buttonwood Sep 2 at 20:34
  • $\begingroup$ @Buttonwood: I agree that it would be nice to compare such a graphic representation. I do not know any such plot for ISOL24 though. $\endgroup$ – user1271772 Sep 3 at 19:48
13
$\begingroup$

This sounds like you were exploring work at least related to the work by the Lilienfeld group equally hosting a dedicated site here about data sets already used in their earlier and ongoing exploration of chemical space, programs used to work with the data, and publications.

To go considerably higher in molecule count than QM9, you could either go for

  • GDB-11 about small organic molecules up to 11 atoms of C, N, O and F which «contains 26.4 million molecules (110.9 million stereoisomers), including three- and four-membered rings and triple bonds», described in J. Chem. Inf. Model. 2007, 47, 342-353 (doi.org/10.1021/ci600423u), or

  • GDB-13, about «small organic molecules up to 13 atoms of C, N, O, S and Cl following simple chemical stability and synthetic feasibility rules. With 977 468 314 structures, GDB-13 is the largest publicly available small organic molecule database to date». This one was described in J. Am. Chem. Soc. 2009, 131, 8732-8733 (doi.org/10.1021/ja902302h)

Convienently, you can download both -- including sub-sets like «containing only carbon and nitrogen», or «chlorine and sulfur», or «fragrance like» in case you don't want to fetch 2GB of already compressed data -- from the Reymond group. To quote: «All the molecules are stored in dearomatized, canonized SMILES format.»

The even larger GDB-17 («of up to 17 atoms of C, N, O, S, and halogens» with an universe of 166 billion entries, described in J. Chem. Inf. Model. 2012, 52, 2864-2875, [doi.org/10.1021/ci300415d, open access]) is accessible to the public on this site as a 50 million random subset only, partly because the gzipped archive is about 400GByte. Among the publications citing this work is for example the Lilienfeld group again for machine learning (J. Chem. Phys. 143, 084111 (2015), doi.org/10.1063/1.4928757).


Initially, I misinterpreted the question but think the answer may be more rounded by addition of the following complementary publication: «Chemical diversity in molecular orbital energy predictions with kernel ridge regression» (J. Chem. Phys. 150, 204121 (2019), doi.org/10.1063/1.5086105, preprint available here). Aiming for a machine-learning analysis, the authors first compared QM9, 44k conformers of proteinogenic amino acids (AA), and a 64k set of organic molecules extracted from the CCDC potentially suitable for organic electronics (OE) for the content of atoms per molecule and found the following distribution:

enter image description here

To shed some light on them:

  • QM9 represents 133,814 small organic molecules with up to 9 heavy atoms (C, N, O and F)
  • AA is about «44,004 isolated and cation-coordinated conformers of 20 proteinogenic amino acids and their amino-methylated and acetylated (capped) dipeptides. The molecular structures are made of up to 39 atoms including H, C, N, O, S, Ca, Sr, Cd, Ba, Hg and Pb.»
  • OE is about «64,710 large organic molecules with up to 174 atoms extracted from organic crystals in the Cambridge Structural Database (CSD). [...] The OE dataset is not yet publicly available. OE offers the largest chemical diversity among the sets in this work both in terms of size as well as number of different elements (Fig. 2). It contains the 16 different element types H, Li, B, C, N, O, F, Si, P, S, Cl, As, Se, Br, Te and I.»

(The mentioned restriction sharing the original data relates to the user agreement with the CCDC.)

Further DFT-based property computations with these OE extracted molecular geometries lead to an ensemble of equilibrium molecular structures, and these derived geometries are accessible within a public Jupyter notebook. Shared with the public here, the deposit comes with a guiding tutorial.ipynb, including an example how to retrieve these optimized geometries and display them with Jmol.

$\endgroup$
  • $\begingroup$ I don't think any of these have more than 81 atoms! $\endgroup$ – user1271772 Sep 1 at 1:31
  • $\begingroup$ Ok I see the confusion. @Buttonwood perhaps you can answer this question: chemistry.stackexchange.com/questions/119804/… $\endgroup$ – user1271772 Sep 1 at 1:32
  • $\begingroup$ @user1271772 I think the updated answer offers both the avenue toward how to access the chemical space of many more (small) molecules (first half) as well as about molecules still considered as small yet more atoms per molecule. Equally I perceive the diversity in the OE dataset regarding «types of molecules, which atoms are connected with each other» larger than for proteïns. For a small-molecule chemist's view, the complexity of the later group emerges more from how theses amino acids are arranged in space ($\beta$-sheets, hairpins, helices, to mention a few complications). $\endgroup$ – Buttonwood Sep 1 at 21:22
  • $\begingroup$ Ok, so now you've shown two databases with fewer than 81 atoms, and one that has more than 81 atoms but is "not yet publicly available". The first half of this answer is completely unrelated to the question so I think it doesn't belong here at all. $\endgroup$ – user1271772 Sep 1 at 23:35
  • $\begingroup$ The first part was based on a misunderstanding, clarified later. The second includes the derived coordinates e.g. in the df_62k.json for which both Jupyter notebook tutorial.ipynb (and equally tutorial.html) document section #4 / working with molecular structures. The optimized molecular structures are saved in xyz_pbe_relaxed. These entries contain molecular coordinates in the standard xyz-format. See on line In [52] as example # We will write the molecules contained in df_subset2 to an xyz-trajectory (stacked xyz files) and these xyz strings are processed further, too. $\endgroup$ – Buttonwood Sep 2 at 20:54
2
$\begingroup$

Beyond other answers, I'd suggest the original PubChemQC project, which offers ~3 million molecules from PubChem optimized using DFT (B3LYP/6-31G*). Molecules include a wide variety of elements as long as the molecular mass is less than 500 Da. (Roughly speaking that should still handle ~38 carbon atoms.)

"PubChemQC Project: A Large-Scale First-Principles Electronic Structure Database for Data-Driven Chemistry" J. Chem. Inf. Model. 2017 57(6) pp. 1300-1308

You mention the number of heavy atoms, but keep in mind that QM9 only contains a small subset of elements and ZINC has many more.

$\endgroup$
  • 1
    $\begingroup$ It's somewhat of a pain to download the PubChemQC data files but they've demonstrated the use in ML methods both for ground-state energies and excitation energies using TD-DFT. $\endgroup$ – Geoff Hutchison Sep 3 at 20:14

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.