👀 mols2grid

mols2grid is an interactive chemical viewer for 2D structures of small molecules, based on RDKit.

➡️ Try the demo notebook on Google Colab, or the more advanced notebook from the RDKit UGM 2021.

🐍 Installation

mols2grid was developped for Python 3.7+ and requires rdkit (>=2020.03.1), pandas and jinja2 as dependencies.
The easiest way to install it is from conda:

conda install -c conda-forge mols2grid

Alternatively, you can also use pip:

pip install mols2grid

It is compatible with Jupyter Notebook and Google Colab (Visual Studio notebooks and Jupyterlab are not fully supported) and can run on Streamlit.

📜 Usage

import mols2grid

mols2grid.display("path/to/molecules.sdf",
                  # RDKit's MolDrawOptions parameters
                  fixedBondLength=25,
                  # rename fields for the output document
                  rename={"SOL": "Solubility",
                          "SOL_classification": "Class",
                          "NAME": "Name"},
                  # set what's displayed on the grid
                  subset=["ID", "img", "Solubility"],
                  # set what's displayed on the tooltips
                  tooltip=["Name", "smiles", "Class", "Solubility"],
                  # style for the grid labels and tooltips
                  style={"Solubility": lambda x: "color: red" if x < -3 else "color: black"},
                  # change the precision and format (or other transformations)
                  transform={"Solubility": lambda x: f"{x:+.2f}"})

Input parameters

You can setup the grid from various inputs:

• a pandas DataFrame (with a column of SMILES or RDKit molecules, controlled by the smiles_col and mol_col parameters),
• a list of RDKit molecules (with properties accessible through the mol.GetPropsAsDict() method),
• or an SDF file (.sdf or .sdf.gz)

You can also rename each field of your input with the rename parameter. Please note that 3 fields are automatically added regardless of your input: mols2grid-id, SMILES and img. If a "SMILES" field already exists, it will not be overwritten.

Parameters for the drawing of each molecule

• useSVG=True: use SVG images or PNG
• coordGen=True: use the coordGen library instead of the RDKit one to depict the molecules in 2D
• removeHs=False: remove explicit hydrogen atoms from the drawings
• size=(160, 120): size of each image
• use_coords=False: use the coordinates of the input molecules if available
• MolDrawOptions=None: RDKit's MolDrawOptions class. Useful for making highly customized drawings. You can also leave this to None, and directly use the attributes of this class as parameters like addStereoAnnotation=True. Additionally, atomColourPalette is made available to modify the color palette if you are not prerendering images.
• prerender=False: Prerender images for the entire dataset, or generate them on-the-fly when needed
• cache_selection=False : Restores the selection from a previous grid with the same name.

Parameters for the grid

You can control the general look of the document through the template argument:

• template="pages" (default) which is displayed above. It integrates nicely with Jupyter notebooks and has a search bar
• template="table", which displays the full list of molecules (no pages). Useful if you ever need to print the full list of molecules on paper (or print to PDF)

Both templates can be configured with the same parameters (a lot of which are CSS declarations). For the pages template, the following parameters are available:

• subset=None: list or None
  Columns to be displayed in each cell of the grid. Each column's value will be displayed from top to bottom in the same order given here. Use "img" for the image of the molecule. Default: all columns (with "img" in first position)
• tooltip=None: list or None
  Columns to be displayed as a tooltip when hovering/clicking on the image of a cell. Use None for no tooltip.
• tooltip_fmt="<strong>{key}</strong>: {value}": str
  Format string of each key/value pair in the tooltip
• tooltip_trigger="click hover": str
  Sequence of triggers for the tooltip: (click, hover, focus)
• tooltip_placement="bottom": str
  Position of the tooltip: auto, top, bottom, left, right
• n_cols=5: int
  Number of columns per page
• n_rows=3 : int
  Number of rows per page
• border="1px solid #cccccc": str
  Styling of the border around each cell (CSS)
• gap=0: int Size of the margin around each cell in px
• fontsize="12pt": str
  Font size of the text displayed in each cell (CSS)
• fontfamily"'DejaVu', sans-serif": str
  Font used for the text in each cell (CSS)
• textalign="center": str
  Alignment of the text in each cell (CSS)
• hover_color="#e7e7e7": str
  Background color when hovering a cell (CSS)
• style=None: dict or None
  CSS styling applied to each item in a cell. The dict must follow a key: function structure where the key must correspond to one of the columns in subset or tooltip. The function takes the item's value as input, and outputs a valid CSS styling. For example, if you want to color the text corresponding to the "Solubility" column in your dataframe:

  style={"Solubility": lambda x: "color: red" if x < -3 else ""}
  

  You can also style a whole cell using __all__ as a key, the corresponding function then has access to all values for each cell:

  style={"__all__": lambda x: "background-color: yellow" if x["Solubility"] < -5 else ""}
  

• transform=None: dict or None
  Functions applied to specific items in all cells. The dict must follow a key: function structure where the key must correspond to one of the columns in subset or tooltip. The function takes the item's value as input and transforms it. For example, to round the "Solubility" to 2 decimals, and display the "Melting point" in Celsius instead of Fahrenheit with a single digit precision and some text before ("MP") and after ("°C") the value:

  transform={"Solubility": lambda x: f"{x:.2f}",
             "Melting point": lambda x: f"MP: {5/9*(x-32):.1f}°C"}
  

  These transformations only affect columns in subset and tooltip and do not interfere with style.
• selection=True : bool
  Enables the selection of molecules using a checkbox. Only usefull in the context of a Jupyter notebook. You can retrieve your selection of molecules (index and SMILES) through mols2grid.get_selection()
• cache_selection=False : bool Restores the selection from a previous grid with the same name if True
• custom_css=None : str or None
  Custom CSS properties applied to the content of the HTML document
• custom_header=None : str or None
  Custom libraries (CSS or JS) to be loaded in the header of the document
• callback=None : str or callable
  JavaScript or Python callback to be executed when clicking on an image. A dictionnary containing the data for the full cell is directly available as data in JS. For Python, the callback function must have data as the first argument to the function. All the values in the data dict are parsed as strings, except "mols2grid-id" which is always an integer.
• sort_by=None : str or None
  Sort the grid according to the following field (which must be present in subset or tooltip).
• substruct_highlight=True : bool Highlight the query when using a SMARTS substructure search. Only available when prerender=False.

Less options are available for the table template, you can check the complete list of arguments with help(mols2grid.MolGrid.to_table)

Output parameters

You can either:

• save the grid with mols2grid.save(input, output="path/grid.html", ...). The file that is generated is a standalone HTML document that should work with most web browsers.
• display it directly in a Jupyter notebook with mols2grid.display(...) (optionnal argument: width="100%", height=None)

🚀 Resources

• Simple exemple by iwatobipen
• Creating a web app with Streamlit for filtering datasets:
  • Blog post by Justin Chavez
  • Video tutorial by Data Professor
• Viewing clustered chemical structures by Pat Walters
• RDKit UGM 2021 notebook

👏 Acknowledgments

• @fredrikw (contributor)
• @JustinChavez (contributor)
• @hadim (conda feedstock maintainer)

⚖ License

Unless otherwise noted, all files in this directory and all subdirectories are distributed under the Apache License, Version 2.0:

    Copyright 2021 Cédric BOUYSSET

    Licensed under the Apache License, Version 2.0 (the "License");
    you may not use this file except in compliance with the License.
    You may obtain a copy of the License at

        http://www.apache.org/licenses/LICENSE-2.0

    Unless required by applicable law or agreed to in writing, software
    distributed under the License is distributed on an "AS IS" BASIS,
    WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
    See the License for the specific language governing permissions and
    limitations under the License.