drugstone 0.2.5

The python package for the https://drugst.one/ platform.

Drugstone

This is the python package for the drugst.one platform.

This package offers tools for drug-repurposing and is a programmatic approach to the functionality of the web portal. For more information visit: https://drugst.one/

Installation

Drugstone depends on a few packages to work. You can use pip to install them.

pip install urllib3 requests pandas pyvis upsetplot

Then you can install drugstone.

pip install drugstone

Finally, it should be possible to import drugstone to your python script.

import drugstone

You can use

import drugstone as ds

to access the complete drugstone API with the ds. notation.

Drugstone officially supports Python 3.6+.

Supported features

Drugstone offers a toolbox for drug repurposing applications.

• Search for drugs, interacting with a list of genes
• Search for drug targets, for a list of genes
• Visualize data in common formats like JSON or CSV
• create interaction graphs for drug and gene interactions

Available Datasets

Protein-protein interactions (ppi_dataset):

NeDRex, BioGRID, IID, IntAct, STRING, APID

Protein-drug interactions (pdi_dataset):

NeDRex, DrugBank, Drug Central, ChEMBL, DGIdb

Please note that some of the datasets require you to accept their terms and conditions before usage. DrugBank can only be used if the license has been agreed to and since NeDRex includes DrugBank data, only a part of NeDRex is available without agreeing to our license.

The terms and conditions can be read by calling

drugstone.print_license()

and can be accepted after reading with

drugstone.accept_license().

Start a new task

With Drugstone it is easy and convenient to search for drugs or drug-targets, starting with a list of genes.

from drugstone import new_task

genes = [
    "CFTR", "TGFB1", "SCNN1B", 
    "DCTN4", "SCNN1A", "SCNN1G",
    "CLCA4", "TNFRSF1A", "FCGR2A"
]

parameters = {
    "target": "drug",
    "algorithm": "trustrank"
}

task = new_task(genes, parameters)

r = task.get_result()

genes = r.get_genes()
drugs = r.get_drugs()

# save directly to files
r.download_json()
r.download_graph()

Start multiple tasks

You can start multiple tasks at once, either with completely independent parameters or with same parameters and different algorithms.

Multiple algorithms

By defining an algorithms value in the parameters dictionary, you can pass a list of algorithm values. For every algorithm, a task will be started, with otherwise same parameter values.

from drugstone import new_tasks

genes = [
    "CFTR", "TGFB1", "SCNN1B", 
    "DCTN4", "SCNN1A", "SCNN1G",
    "CLCA4", "TNFRSF1A", "FCGR2A"
]

parameters = {
    "target": "drug",
    "algorithms": ["trustrank", "closeness", "degree"]
}

tasks = new_tasks(genes, parameters)   
r = tasks.get_result()                 
r.download_json()

Independent parameters

new_tasks() accepts a list of parameter dictionaries. For every dictionary a task will be started.

from drugstone import new_tasks

genes = [
    "CFTR", "TGFB1", "SCNN1B", 
    "DCTN4", "SCNN1A", "SCNN1G",
    "CLCA4", "TNFRSF1A", "FCGR2A"
]

p1 = {
    "target": "drug",
    "ppi_dataset": 'nedrex',
    "pdi_dataset": "drugbank"
}

p2 = {
    "target": "drug",
    "ppi_dataset": 'IID',
    "pdi_dataset": "chembl"
}

p3 = {
    "target": "drug",
    "ppi_dataset": 'apid',
    "pdi_dataset": "dgidb"
}

tasks = new_tasks(genes, [p1, p2, p3]) 
r = tasks.get_result()                 
r.download_json()

Union and intersection of tasks

You can get the union or intersection of tasks. That returns a TaskResult with the according result.

from drugstone import new_tasks

genes = [
    "CFTR", "TGFB1", "SCNN1B", 
    "DCTN4", "SCNN1A", "SCNN1G",
    "CLCA4", "TNFRSF1A", "FCGR2A"
]

parameters = {
    "target": "drug",
    "algorithms": ["trustrank", "closeness", "degree"]
}

tasks = new_tasks(genes, parameters)    

u = tasks.get_union()                  
u.download_json()

i = tasks.get_intersection()
i.download_json()

Combine a drug-target search with a drug search

This will perform a drug-target search for the seed genes and then use the drug-target search results and the seed genes to perform a drug-search. Finally, a Task with the drug-search results will be returned.

from drugstone import deep_search

genes = [
    "CFTR", "TGFB1", "SCNN1B", 
    "DCTN4", "SCNN1A", "SCNN1G",
    "CLCA4", "TNFRSF1A", "FCGR2A"
]

parameters = {
    "algorithm": "trustrank"
}

task = deep_search(genes, parameters)  
r = task.get_result()                  
r.download_json()

Available Parameters

parameters = {
    "algorithm": "trustrank" | "multisteiner" | "keypathwayminer" | "closeness" | "degree" | "proximity" | "betweenness",
    "ppi_dataset": "NeDRex",
    "pdi_dataset: "NeDRex",
    "result_size": 20,
    "target": "drug" | "drug-target",
    "include_indirect_drugs": True | False,
    "include_non_approved_drugs": True | False,
    "max_deg": sys.maxsize, # filter out nodes with high degrees
    "hub_penalty": 0.0, # penalize hub nodes
    "filter_paths": True | False, # include only shortest connections in the result

    "damping_factor": 0.85, # only in trustrank

    "num_trees": 5, # only in multisteiner
    "tolerance": 10, # only in multisteiner

    "k": 5, # only in keypathwayminer
} 

For more information about the algorithms, please refer to https://drugst.one/doc#implementation_algorithms.

For more information abouyt the available dataset types, please refer to https://drugst.one/doc#implementation_datasources.

class Task

Represents a task.

get_result() -> TaskResult
Returns a TaskResult for the result of the task.

get_info() -> dict
Returns a dict with information about the task.

get_parameters() -> dict
Returns a dict with the parameters of the task.

class TaskResult

Represents the results of a task.

get_genes() -> dict
Returns a dict with the genes.

get_drugs() -> dict
Returns a dict with the drugs.

to_dict() -> dict
Returns a dict with the result.

to_pandas_dataframe() -> DataFrame
Returns a pandas DataFrame of the result.

download_json(path: str, name: str) -> None
Downloads a json file with the result.

download_genes_csv(path: str, name: str) -> None
Downloads a csv file with the genes of the result.

download_drugs_csv(path: str, name: str) -> None
Downloads a csv file with the drugs of the result.

download_edges_csv(path: str, name: str) -> None
Downloads a csv file with the edges of the result.

download_graph(path: str, name: str) -> None
Downloads a html file with a graph of the nodes.

class Tasks

Wraps a list of Task objects.

get_result() -> TasksResult
Returns a TasksResult for the list of tasks.

get_union() -> TaskResult
Returns a TaskResult with the union of the tasks.

get_intersection() -> TaskResult
Returns a TaskResult with the intersection of the tasks.

class TasksResult

Represents the results of a list of Task objects.

get_tasks_list() -> List[Task]
Returns the list of tasks.

to_dict() -> dict
Returns a dict with the results of the tasks.

download_json(path: str, name: str) -> None
Downloads a json file with the results.

create_upset_plot() -> None
Opens a new window with an upset plot of the results.

Copyright: 2022 - Institute for Computational Systems Biology by Prof. Dr. Jan Baumbach
Author: Ugur Turhan