suggestion: more restrictive cation/anion definitions

[wenyan4work]

The current definition is

        cation: str = "[+{1-},$([NX3&!$([NX3]-O)]-[C]=[NX3+])]",
        anion: str = "[-{1-},$(O=[C,S,P]-[O-])]",

I suggest a slightly more restrictive definition:

        cation: str = "[+{1-}!$(*~[*-]),$([NX3&!$([NX3]-O)]-[C]=[NX3+])]",
        anion: str = "[-{1-}!$(*~[*+]),$(O=[C,S,P]-[O-])]",

such that an cation atom is defined as an atom with positive formal charge but not immediately connected to an atom with a negative formal charge.

For example, dmso and azides are sometimes written as

C[S+](-[O-])C
R-N=[N+]=[N-]

These are probably not what we are looking for when searching for 'ionic' interactions.

[cbouy]

Hi @wenyan4work and thanks for opening a detailed issue (and a PR!)

I agree that DMSO shouldn't be labelled as involved in ionic interactions, but this sounds more like an issue with structure standardisation than with ProLIF's interaction definitions.
For azides, I'm a bit more nuanced, I wouldn't be surprised seeing it labelled with an Ionic interaction, but the charge delocalisation would make it more of a dipole-ion interaction than a typical salt-bridge. There isn't such interaction in ProLIF though, and Ionic is the closest match, so I'm not sure if it makes sense to remove it or not.
So it's unlikely that I'll accept the !$(*~[*-])/!$(*~[*+]) changes as is, but we could potentially add an azide exclusion. What do you think?

[wenyan4work]

Hi @cbouy , the DMSO is an example but the problem is more general than DMSO. The nitro functional group (and similar NX5 functional groups) is also often represented as -[N+](=O)[O-]. Pyridine-N-oxide is a less common but more subtle example: c1cc[n+](cc1)[O-]. Nitryl fluoride also shows this problem but it is rarely found in PL complex so probably we can ignore it.

Since widely used structure sanitization and standardization tools (like rdkit) produce a lot of such [X+]-[Y-] structures, I do suggest a more restrictive description of ionic interactions.