Mol2 files bonds

[btyukodi]

PandasMol2().read_mol2() reads and parses a mol2 file, however, in the dataframe, only the @<TRIPOS>ATOM section is present. Is there any way to access the bonds?

Thanks Botond

[rasbt]

Hi there,

in the current version of biopandas, the Mol2 BOND section is not parsed, yet. You would need to parse it from the raw text data that is attached to a PandasMol2 object pmol via pmol.mol2_text.

I think adding some parsing functionality for the BOND section would be a good thing to do!

---

Since the analysis of bonds is not necessarily done in each virtual screening analysis/application, I would make parsing the bonds section not the default (but optional) for computational efficiency reasons.

A general "design" question now is whether a BOND DataFrame should be stored/accessed via

a) pmol.df_bonds (where pmol.df is currently storing the atom section)

or

b) pmol.df['bonds'] (where pmol.df would then have to be renamed to pmol.df['atom'], which would make it more consistent with PandasPdb())

pmol.parse_bonds() # method to parse the bonds section
pmol.df_bonds # dataframe containing the bonds section

Feedback and comments (and PRs!) are welcome!

[btyukodi]

Hi, thanks for your answer. Just wrote a quick, dirty and probably non-optimal parser. It could serve as a temporary solution:

def bond_parser(filename):
    f = open(filename,'r')
    f_text = f.read()
    f.close()
    bond_start = f_text.find('@<TRIPOS>BOND')
    bond_end = f_text[bond_start:].replace('@<TRIPOS>BOND','').find('@')
    df_bonds = pd.DataFrame(np.array(f_text[bond_start:bond_end].replace('@<TRIPOS>BOND\n','').replace('\n',' ').split(' ')).reshape((-1,4)),
            columns=['bond_id', 'atom1', 'atom2', 'bond_type'])
    df_bonds.set_index(['bond_id'], inplace=True)
    return df_bonds

[wlgfour]

Here is a shorter version of of what @btyukodi wrote that uses regex:

dwef bond_parset(filename):
    with open(filename, 'r') as f:
        f_text = f.read()
    bonds =  np.array(re.sub(r'\s+', ' ', re.search(r'@<TRIPOS>BOND([a-z0-9\s]*)@', f_text).group(1)).split()).reshape((-1, 4))
    df_bonds = pd.DataFrame(bonds, columns=['bond_id', 'atom1', 'atom2', 'bond_type'])
    df_bonds.set_index(['bond_id'], inplace=True)
    return df_bonds

[rasbt]

Thanks a lot. I totally forgot about this / got to busy to work on this. Thanks for sharing though, I hopefully will get to it one day (or maybe someone else :))

[mercicle]

Hey there all, was this ever implemented?

[rasbt]

No, sorry, never really found the time to do that :(

[Kroppeb]

Seems like this is still an open issue?

[rasbt]

yeah, this is still an open issue