assign bond by comparison with covalent radii #8
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Test-wise addition for / about nitrogen. Tested on the example of pyridine by openbabel from the SMILES string (`c1ccncc1`), and diethylamine (`CCNCC`).
Tested for the examples of diethylether (`CCOCC`), THF (`C1COCC1`), and propan-2-ol (`C(C)(O)(C)`) in .xyz files generated by obabel.
Tested with openbabel generated .xyz about diethyl thioether (`CCSCC`) and furfural (`O=Cc1cccs1`). However, contrasting to e.g., the depiction with openbabel generated .pov, the C-S bonds no longer are visible.
Tested with openbabel SMILES string of fluoroethane (`CCF`). The .pov/.ini and subsequent generation of .png seem fine.
Introduction of chlorine. Tested with an .xyz about chloroethane (`CCCl`) generated by openbabel, the chlorine atom is depicted as sphere, separated from the backbone of the molecule. Because of a similar observation for compounds with C-S bonds, and because the analogue CCF was rendered well, perhaps the assignment where to draw a bond faces an obstacle here.
The previous approach requires four lines each for every atom type as an instance of class Atom. I think an approach which uses the element symbol as key to a the corresponding values of mass, radiosity, and rgb coordinates in a dictionary is - in comparison to this one - easier to read, and to maintain. Lacking any better inspiration, atoms with atom types not yet explicitly defined in said dictionary default to hydrogen. In such an instance, the conversion will not stop for good, however is going to issue one warning for each atom not fully defined (yet).
The previous definition of bonds if distances are less than 1.6 A is too simple. For one, there are covalent bonds which exceed this threshold (C-I). For two, simple increase of this limit to say 1.8 A can cause geminal hydrogens on methylene units to be assigned a H-H bond in addition to an already existing C-H bond. This commit ventures out to add the property dictionary of the atoms additional values of covalent radii, and the standard deviations of these covalent radii. Eventually, there shall be a comparison between the distance as computed reading the .xyz file, and a theoretical threshold then used to assign a bond, or not. For now, the implementation is after the 1.6 A limit and only considers C and H as test subjects. Hence, the test is not yet as effective as wanted.
Though still incomplete, it is time to move the currently relevant operations into a function.
Instead of a fix 1.6 Angstrom limit applicable on all bonds as a maximal bond length, the check to build one bond, or not now is based on comparing covalent radii and distances inferred from the xyz file read.
In previous versions of the script, there was the uniform threshold of interatomic distances to check if two atoms could form a bond (any kind of covalent bond). With 1.6 A, this was fine for C-C, and C-H, but too short e.g. for C-S, or C-I. An increase to 1.8 A rendered the display of the longer bonds possible; however, simultaneously built H-H bonds for geminal hydrogens on alkyl groups. To test the new approach based on comparison of covalence radii, benzene.xyz now is complemented by pentylbenzene.xyz to extend the coverage of tests possible.
The test with pentylbenzene.xyz is encouraging (no spooky H-H bonds for geminal H on alkyl chains), and the greeter from the test function can be removed.
The coverage of the script's dictionary and covalent radii is extended, added are N, O, and S.
The coverage of atomic radii is extended -- two periods H to Ar, plus the more frequently encountered Br and I. More as a proof of 'a priori, the concept is working', than backed by chemical evidence for such a compound, an example with C, H, N, O, P, S, and I is added as a colorful test model. Coverage should be enough for many simpler molecular compounds. Lines of the chemical dictionary incomplete for covalent radii are currently muted which doesn't affect the output of the other elements.
The atom radii are covered all between number 1 (hydrogen, H) and 96 (Curium, Cm). This likely covers most cases of application of this script in mind.
The sequence of presenting the covalent radii, their standard deviation; the results of subsequent computation of a threshold to decide if a covalent bond could form (based on atoms assumed as uncharged) was changed. Explicit format definitions on the fixed precision.
The user now has to press -v (or --verbose) to obtain a detailed log to the CLI.
Contributor
Author
|
To ease a review, individual commits are retained separate. I'm then can squash-merge some of them. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
In previous commits, the assignment of bonds was based on a universal
distance threshold. By the present approach, the atom dictionary was
extended by tabulated data about covalent atomic radii and standard
deviations of these data which are used to decide if two atoms assumed
as neutral are close enough to form a bond. The individual decision
can be reported to the CLI by the optional
-v, or--verboseflag.This commit builds on top of an PR,[1] and addresses an issue.[2]
[1] #7
[2] #6