Home

Welcome to the CIDER wiki!

How to use the notebook

Import datasets

The first step after the set-up is to load the SDFiles for comparison into the notebook with the 'import_data_as_dict' function. The function will create a dictionary which content will be updated with every subsequent function used. This main dictionary contains a sub-dictionary for every SDFile named after the SDFile name.

function	import_data_as_dict
parameter	path_to_data : str
	Path to the SDFiles which are to be compared.

In the Jupyter Notebook the files from 'data2' are imported but one can also change the import function to 'data'. Using the SDFiles from 'data', one can see that the functions are also working on datasets of different lenght and how the intersection in the Venn Diagramm between all three datasets looks like.

Overview

To get an overview of the datasets, the number of molecules in every dataset can be determined using the 'get_number_of_molecules' function. The results for every dataset are included in the dataframe.

function	get_number_of_molecules
parameter	all_dict : dict
	Name of the dictionary with the imported SDFiles.

Additionally, one can create a grit image of molecules from each dataset with 'draw_molecules' function. The number of displayed molecules from every dataset can be specified (parameter: number_of_mols) as well as the size of the images (parameter: image_size). The images will also be saved in an output folder in a chosen format (parameter: data_type).

function	draw_molecules
parameter	all_dict : dict
	Name of the dictionary with the imported SDFiles.
	number_of_mols : int, default=12
	Number of molecules from every dataset to be displayed.
	image_size : int, default=200
	Image size for one molecule in the grid image.
	data_type : str, default='png'
	Data type for the exported image.

Example of the 'draw_molecules' function with the visualization of set_phenole.sdf as grit image (number_of_molecules = 6).

Identifier

For the subsequent comparison, the molecules need a string representation and therefore one can use SMILES, InChI or InChIKey strings. The 'get_identifier_list_key' function gets the chosen identifier strings (parameter: id_type) for every molecule and puts them in the dataframe.

function	get_identifier_list_key
parameter	all_dict : dict
	Name of the dictionary with the imported SDFiles.
	id_type : str, default='inchi'
	Type of Identifier ("inchi", "inchikey" or "smiles").

If one uses datasets from a certain database and the database provides its own ID, this dataset ID can also be extracted from the SDFiles and stored in the dataframe. Therefore, one can use the function 'get_database_id' and input the ID name (parameter: id_name).

function	get_database_id
parameter	all_dict : dict
	Name of the dictionary with the imported SDFiles.
	id_name : str
	Name of the database ID in the SDFiles.

Molecule comparison

With 'get_shared_molecules_key' one gets the number and identifier string for those molecules which are present in all of the compared datasets.

function	get_shared_molecules_key
parameter	all_dict : dict
	Name of the dictionary with the imported SDFiles.

If there are no more than three datasets, a Venn diagram of the intersection of the datasets can be created using 'visualize_intersection'. The image will be saved in the output folder in a chosen format (parameter: data_type).

function	visualize_intersection
parameter	all_dict : dict
	Name of the dictionary with the imported SDFiles.
	data_type : str, default='png'
	Data type for the exported image.

Example of an intersection from three compared datasets created with the 'visualize_intersection' function.

Descriptor and descriptor value distribution

The function 'get_descriptor_list_key' utilizes a callable function (parameter: descriptor) from RDKit to get descriptor values for every molecule. For example, the callable functions can be from the rdkit.Chem.rdMolDescriptors module or the rdkit.Chem.Descriptors module. The values are saved in the dataframe under a chosen name (parameter: descriptor_list_keyname).

function	get_descriptor_list_key
parameter	all_dict : dict
	Name of the dictionary with the imported SDFiles.
	descriptor : callable
	RDKit function returning a molecular descriptor value.
	descriptor_list_keyname : str
	Name for referring to descriptor values.

To get and visualize the distribution of the descriptor values, the function 'descriptor_counts_and_plot' can be used. The function distinguishes between continuous and discrete distributed descriptor values and for the continuous values one can choose the binning size (parameter: width_of_bins). The distribution of values is exported as a csv-file and the visualization with a selectable format (parameter: data_type) is also saved in the output folder.

function	descriptor_counts_and_plot
parameter	all_dict : dict
	Name of the dictionary with the imported SDFiles.
	descriptor_list_keyname : str
	Name referring to descriptor values which are to be visualized.
	width_of_bins : int, default=10
	Interval size for binning of continuous descriptor values.
	data_type : str, default='png'
	Data type for the exported image.
	save_dataframe : bool, default=True
	Option to export the counts of the descriptor values as csv.

Example of the a descriptor value distribution from the 'descriptor_counts_and_plot' function. The LogP values of the three datasets are binned in intervals of 5.

With the database ID one can also search for the descriptor value of a specific molecule using the 'get_value_from_id' function. The function tells in which SDFile the molecule is found and the value of the descriptor.

function	get_value_from_id
parameter	all_dict : dict
	Name of the dictionary with the imported SDFiles.
	wanted_id : str
	Database ID from the molecule in question.
	descriptor_list_keyname : str
	Name referring to descriptor values which are to be visualized.

Lipinski Rule of 5

With 'get_lipinski_key' the number of broken Lipinki Rules for every molecule is calculated and a summary for every SDFile is created.

function	get_lipinski_key
parameter	all_dict : dict
	Name of the dictionary with the imported SDFiles.

Subsequently, the 'lipinski_plot' function visualizes the number of broken rules. Again, the results are exported as a csv-file and the bar-plot is also saved with a selectable format (parameter: data_type) in the output folder.

function	lipinski_plot
parameter	all_dict : dict
	Name of the dictionary with the imported SDFiles.
	data_type : str, default='png'
	Data type for the exported image.
	save_dataframe : bool, default=True
	Option to export the counts of the descriptor values as csv.

Example of a Lipinski Plot with three datasets created with the 'lipinski_plot' function.

Chemical Space Visualization

For the visualization of the chemical space, the chemplot module is used. The extended connectivity fingerprints can be specified with the fingerprint radius (parameter: fp_radius) and the size (parameter: fp_bits). For the dimension reduction PCA, t-SNE or UMAP can be chosen (parameter: dimension_reduction). The chemical space plot is saved in the output folder except when choosing to create an interactive plot (parameter: interactive). Then the plot will be displayed and can be manually saved.

function	chemical_space_visualization
parameter	all_dict : dict
	Name of the dictionary with the imported SDFiles.
	fp_radius : int, default=2
	Radius of the Extended Connectivity Fingerprints.
	fp_bits : int, default=2048
	Size of the Extended Connetivity Fingerprints.
	dimension_reduction : str, default='pca'
	Method of dimension reduction ("pca", "umap" or "tsne").
	interactive : bool, default=True
	Option to create an interactive plot.

Example of the function 'chemical_space_visualization' with three datasets (dimension_reduction='tsne', interactive=False).

Export

The calculated descriptor values for every molecule can exported as a csv-file using the 'export_single_dict_values' function. For every imported dataset there will be a separate export file containing the values.

function	export_single_dict_values
parameter	all_dict : dict
	Name of the dictionary with the imported SDFiles.

Additionally, a summary of all the created images in the form of a pdf with all images can be created with the 'export_all_picture_pdf' function. This file will not include images that are saved as pdf beforehand.