Substructure Search with Pandas and OENotebook

Substructure searching is one of the fundamental tools of cheminformatics. In this example we'll use the substructure searching in OEChem in conjuction with the pandas dataframes that OENotebook provides.

In [2]:
from __future__ import print_function

# import OE tools and data
import oenotebook as oenb
In [3]:
# Read molecules into a Pandas Dataframe
df = oenb.read_file_to_dataframe("./eMol_ran50.ism", title_col="Title")

# Check to make sure data looks right
Molecule Title
0 25864246
1 25167497
2 24665960
3 2251284
4 5632287

We'll create a simple substructure search that looks for a benzyl ring

In [4]:
from openeye import oechem
mol = oechem.OEGraphMol( ) 

# Have some toluene
oechem.OEParseSmiles(mol, "c1ccccc1C")

# create a substructure search object for benzene
ss = oechem.OESubSearch("c1ccccc1") 

if ss.SingleMatch(mol):
    print("benzene matches toluene") 
    print("benzene does not match toluene")

benzene matches toluene

Substructure Search and a DataFrame

Now we can work with this substructure search by generating a new column in our dataframe with the binary output of the substructure match.

In [5]:
# Apply our search to the molecules in the dataframe
df["HasBenzene"] = df.Molecule.apply(ss.SingleMatch)

Molecule Title HasBenzene
0 25864246 True
1 25167497 True
2 24665960 True
3 2251284 False
4 5632287 True
In [14]:
benzyls = df[df.HasBenzene == True]
Molecule      44
Title         44
HasBenzene    44
dtype: int64

Rendering the DataFrame with a Custom Depiction

Finally we'll write a custom depiction function (derived from this OEDepict example ) for our substructure search to highlight the matches for each molecule in our dataframe. Once we've written and confirmed this function depicts the match, we'll pass it to the render_dataframe function to see our custom depiction in the dataframe.

In [26]:
from openeye import oedepict
from IPython.display import HTML, display
from base64 import b64encode

def draw_subsearch_highlights(mol, width=400., height=400., subsearch=ss):
    opts = oedepict.OE2DMolDisplayOptions(width, height, oedepict.OEScale_AutoScale)

    mol = oechem.OEGraphMol(mol)
    img = oedepict.OEImage(width, height)
    hstyle = oedepict.OEHighlightByBallAndStick(oechem.OEBlueTint)

    disp = oedepict.OE2DMolDisplay(mol, opts)
    unique = True
    for match in subsearch.Match(mol, unique):
        oedepict.OEAddHighlighting(disp, hstyle, match)
    oedepict.OERenderMolecule(img, disp)
    return oenb.draw_oeimage_to_img_tag(img)
In [30]:
In [31]:
oenb.render_dataframe(df.head(),render_func=lambda x, width, height: draw_subsearch_highlights(x, width, height))
Molecule Title HasBenzene
0 25864246 True
1 25167497 True
2 24665960 True
3 2251284 False
4 5632287 True

This post was written in a Jupyter Notebook. You can find the notebook here with instructions for downloading and running it yourself.