Using the ACS1996 drawing style in PandasTools

Making small structure drawings more legible

This is a short one showing how to improve the legibility of small structure drawings in Pandas DataFrames displayed in the notebook.

from rdkit import Chem
from rdkit.Chem import Draw
from rdkit.Chem.Draw import IPythonConsole
from rdkit.Chem import PandasTools
PandasTools.RenderImagesInAllDataFrames(True)

Start by reading in an SDF:

df = PandasTools.LoadSDF('../data/chembl26_very_active.sdf.gz')
df.head()

[04:36:24] Warning: ambiguous stereochemistry - zero final chiral volume - at atom 50 ignored
[04:36:25] Warning: ambiguous stereochemistry - overlapping neighbors  - at atom 7 ignored
[04:36:25] Warning: ambiguous stereochemistry - overlapping neighbors  - at atom 9 ignored
[04:36:25] Warning: ambiguous stereochemistry - overlapping neighbors  - at atom 9 ignored

	compound_chembl_id	assay_chembl_id	target_chembl_id	pref_name	standard_relation	standard_value	standard_units	standard_type	ID	ROMol
0	CHEMBL86971	CHEMBL690412	CHEMBL243	Human immunodeficiency virus type 1 protease	=	0.8	nM	Ki
1	CHEMBL6710	CHEMBL621508	CHEMBL324	Serotonin 2c (5-HT2c) receptor	=	0.26	nM	Ki
2	CHEMBL314218	CHEMBL677606	CHEMBL248	Leukocyte elastase	=	0.03	nM	Ki
3	CHEMBL438897	CHEMBL877834	CHEMBL1855	Gonadotropin-releasing hormone receptor	=	0.24	nM	Ki
4	CHEMBL15928	CHEMBL616937	CHEMBL1983	Serotonin 1d (5-HT1d) receptor	=	0.7	nM	Ki

Move the molecule to the first column

cols = list(df.columns)
cols.insert(0,cols.pop(-1))
df = df[cols]
df.head()

	ROMol	compound_chembl_id	assay_chembl_id	target_chembl_id	pref_name	standard_relation	standard_value	standard_units	standard_type	ID
0		CHEMBL86971	CHEMBL690412	CHEMBL243	Human immunodeficiency virus type 1 protease	=	0.8	nM	Ki
1		CHEMBL6710	CHEMBL621508	CHEMBL324	Serotonin 2c (5-HT2c) receptor	=	0.26	nM	Ki
2		CHEMBL314218	CHEMBL677606	CHEMBL248	Leukocyte elastase	=	0.03	nM	Ki
3		CHEMBL438897	CHEMBL877834	CHEMBL1855	Gonadotropin-releasing hormone receptor	=	0.24	nM	Ki
4		CHEMBL15928	CHEMBL616937	CHEMBL1983	Serotonin 1d (5-HT1d) receptor	=	0.7	nM	Ki

The default drawing style doesn’t look great when the molecule images are small like this; the ACS1996 style really works a lot better.

We can use the ACS996 style in the DataFrame by setting the drawing options for PandasTools. We do this by creating a new drawing options instance, setting that to ACS1996 mode, and then setting the drawOptions variable in the PandasTools module.

dopts = Draw.MolDrawOptions()
Draw.SetACS1996Mode(dopts,Draw.MeanBondLength(df.ROMol[0]))

PandasTools.drawOptions = dopts

The new options are now used when we show the DataFrame:

df.head()

	ROMol	compound_chembl_id	assay_chembl_id	target_chembl_id	pref_name	standard_relation	standard_value	standard_units	standard_type	ID
0		CHEMBL86971	CHEMBL690412	CHEMBL243	Human immunodeficiency virus type 1 protease	=	0.8	nM	Ki
1		CHEMBL6710	CHEMBL621508	CHEMBL324	Serotonin 2c (5-HT2c) receptor	=	0.26	nM	Ki
2		CHEMBL314218	CHEMBL677606	CHEMBL248	Leukocyte elastase	=	0.03	nM	Ki
3		CHEMBL438897	CHEMBL877834	CHEMBL1855	Gonadotropin-releasing hormone receptor	=	0.24	nM	Ki
4		CHEMBL15928	CHEMBL616937	CHEMBL1983	Serotonin 1d (5-HT1d) receptor	=	0.7	nM	Ki

The ACS1996 style will also be used when we do things like generate an XLSX file:

PandasTools.SaveXlsxFromFrame(df.head(100),'../data/output.xlsx')

As an aside: we can also set ACS1996 mode for the rendering in the notebook;

Draw.SetACS1996Mode(IPythonConsole.drawOptions,Draw.MeanBondLength(df.ROMol[0]))

df.ROMol[0]

That style is also used in Draw.MolsToGridImage():

Draw.MolsToGridImage(df.ROMol[:8],molsPerRow=4)