Browser Compatibility
OS Version Chrome Firefox Microsoft Edge Safari
Linux Ubuntu 22.04.1 LTS 107.0.5304.87 106.0.5 Not Tested N/A
MacOS Ventura 13.0 107.0.5304.110 106.0.5 Not Tested 16.1
Windows Windows 11 107.0.5304.107 106.0.5 107.0.1418.42 N/A
For best browsing experience, it is strongly recommended to use Chrome. If you have page display problems (such as incomplete pages, content obscured or misplaced) during browsing, try adjusting the browser zoom size.
Introduction
The phenotype of drug action, including therapeutic actions (here, represented by the level III ATCs) and adverse drug reactions (ADRs), is an important indicator for evaluating the druggability of new drug candidates. However, the current calculation methods for predicting the drug action phenotype are mostly concentrated on a single aspect, which can not evaluate the drug safety while clarifying the drug efficacy. In addition, the overall structure similarity strategy of drugs or their related proteins often leads to the inconsistency between the structure and function of drugs, which limits the prediction space of drug actions and also fails to provide structural information inducing serious adverse reactions (SADRs) for optimization. Therefore, there is an urgent need for a new drug candidate screening platform that can simultaneously predict its therapeutic effects and adverse reactions, so as to identify the lead compound at an early stage.
We provide DAPredict, a user-friendly database, in which our original algorithm is embedded, which is based on the local structure of drugs and proteins (see the citation for details) to predict the comprehensive action phenotypes of drugs.
DAPredict mainly provides the following services:
1. The search function provides predicted and manually verified action phenotypes of 1748 approved drugs (see browse drugs on the home page) for searching directly. Users can input a drug (name or ID) or an action phenotype to search for their relationships. More importantly, users can enter a drug and the interest action phenotype at the same time to search for phenotype-corresponding substructures and phenotype-corresponding domains, which are the most important information in structural optimization for eliminating SADRs and exploring potential mechanisms of ADRs or ATCs.
2. The online prediction tool generalizes the algorithm to predict the action phenotypes for most compounds recorded in PubChem (about 110,000,000 in total, see predictable compounds on the home page). These compounds cover almost all the current natural products (168360/168756), and the biological activities of most of them have not been discovered yet due to the lack of prediction methods for comprehensive phenotypes of drug actions. Users only need to use the PubChem ID of a compound as input to quickly predict the relationship between the compound and the action phenotypes online. Similar to approved drugs, users can also explore the mechanism of drug action phenotypes and obtain the phenotype-corresponding substructure information to optimize the drug structures.
3. In the “download” section, our database also provides valuable downloadable resources, including predicted relationships by our original algorithm, and data collated from known relationships from related databases.
All results are presented in interactive graphs and tables and are downloadable. Users can choose how the results are presented and customize the download content.
Our database has the following innovations:
1. it can provide comprehensive phenotypes of drugs with the related score.
2. it overcomes the limitation of traditional drug discovery based on drug phenotype that the mechanism of drug action cannot be known.
3. the original algorithm for predicting drug action phenotype based on the local active structure can provide the local active structure of serious adverse reactions for drug structure optimization.
In a word, it is the first time to provide an easy-to-use platform for predicting drug action phenotypes embedded with our original algorithm. It not only includes the search of action phenotypes of 1748 approved drugs, but also includes the prediction service of about 110,000,000 compounds (covering 99.8% of the currently discovered natural products) recorded in PubChem. Our database would provide pharmaceutical developers and researchers with reliable and broad resources.
Tool
Here, we provide an online tool which is embedded with our new predictive algorithm. The tool relies on compound substructures to predict action phenotypes. Currently, PubChem provides substructure information of more than 110,000,000 compounds, so the tool has broad applicability and can provide great reference value for drug developers. More importantly, based on the substructure of compounds, users can also explore the mechanism of compound action phenotypes.
Compound-ADR
In this section, users can predict relationships between compounds and adverse reactions online.
Input: PubChem CID of compound (fuzzy query is not supported)
Output:
CID: PubChem CID of the compound.
Name: Generic name of the compound.
Formula: Molecular formula of the compound.
ADR Term: ADR term from SIDER.
Score: Predicted relevance score.
Rank: The ranking is based on all the Drug-ADR relationships in the database. Every 20% is a grade, and there are five grades in total.
Compound-ATC
In this section, users can predict relationships between compounds and anatomical therapeutic chemical online.
Input: PubChem CID of compound (fuzzy query is not supported)
Output:
CID: PubChem CID of the compound.
Name: Generic name of the compound.
Formula: Molecular formula of the compound.
ATC Code: ATC (Anatomical Therapeutic Chemical) code (third level).
ATC Description: ATC term corresponding to ATC code.
Score: Predicted relevance score.
Rank: The ranking is based on all the Drug-ATC relationships in the database. Every 20% is a grade, and there are five grades in total.
Compound-Domain-ADR
In this section, users can explore the underlying mechanisms of compound-induced adverse reaction from the target perspective.
Input 1 (compound): PubChem CID of the compound.
Input 2 (ADR): ADR term (not case sensitive, fuzzy query is not supported)
Output:
CID: PubChem CID of the compound.
Name: Generic name of the compound.
Formula: Molecular formula of the compound.
ADR Term: ADR term from SIDER.
Target Gene: Official gene symbol of predicted target gene of compound.
Domain: Domain ID of predicted target protein from Pfam.
Score: Predicted relevance score.
Rank: The ranking is based on all the Drug-Domain-ADR relationships in the database. Every 20% is a grade, and there are five grades in total.
Compound-Domain-ATC
In this section, users can explore the underlying mechanisms of therapeutic effects of compounds from the target perspective.
Input 1 (compound): PubChem CID of the compound.
Input 2 (ATC): ATC code (third level, fuzzy query is not supported)
Output:
CID: PubChem CID of the compound.
Name: Generic name of the compound.
Formula: Molecular formula of the compound.
ATC Code: ATC (Anatomical Therapeutic Chemical) code (third level).
ATC Description: ATC term corresponding to ATC code.
Target Gene: Official gene symbol of predicted target gene of compound.
Domain: Domain ID of predicted target protein from Pfam.
Score: Predicted relevance score.
Rank: The ranking is based on all the Drug-Domain-ATC relationships in the database. Every 20% is a grade, and there are five grades in total.
Compound-Substructure-ADR
In this section, users can explore the underlying mechanisms of compound-induced adverse reaction from the substructure perspective.
Input 1 (compound): PubChem CID of compound (fuzzy query is not supported)
Input 2 (ADR): ADR term (not case sensitive, fuzzy query is not supported)
Output:
CID: PubChem CID of the compound.
Name: Generic name of the compound.
Formula: Molecular formula of the compound.
Bit: The position of the substructure in 881 substructures from PubChem.
Substructure: The chemical structural formula of the substructure.
ADR Term: ADR term from SIDER.
Score: Predicted relevance score.
Rank: The ranking is based on all the ADR-SUB relationships in the database. Every 20% is a grade, and there are five grades in total.
Compound-Substructure-ATC
In this section, users can explore the underlying mechanisms of therapeutic effects of compounds from the substructure perspective.
Input 1 (compound): PubChem CID of compound (fuzzy query is not supported)
Input 2 (ATC): ATC code (third level, fuzzy query is not supported)
Output:
CID: PubChem CID of the compound.
Name: Generic name of the compound.
Formula: Molecular formula of the compound.
Bit: The position of the substructure in 881 substructures from PubChem.
Substructure: The chemical structural formula of the substructure.
ATC Code: ATC (Anatomical Therapeutic Chemical) code (third level).
ATC Description: ATC term corresponding to ATC code.
Score: Predicted relevance score.
Rank: The ranking is based on all the ATC-SUB relationships in the database. Every 20% is a grade, and there are five grades in total.
Visualization
All results are presented in interactive graphs and tables, and users can freely choose how the results are presented and customize the download content.
Interactive Graph
① Sliding window: Users can drag the blue handles at both ends to adjust the size of the sliding window. Drag the sliding window to further explore a specific small range of data.
② Save: Users can save the current graph in png format.
③ Restore: Users can restore the graph to its original form.
④ Bar plot: Users can click this button to switch the default line chart to a bar chart. When the score is very small, it is more recommended to view in the bar chart.
⑤ Rank: The predicted relevance score is divided into five grades (see the search and tool sections for details), and users can click the label to filter the results by rank attribute.
⑥ Interactive nodes: Users can hover over nodes to view details.
Interactive Form
① Search: search within the form is available to further filter valuable results.
② Format conversion: users can switch the present format of result form.
③ Column display: users can customize the displayed columns.
④ Export: users can export the result on the current page in csv or Excel formats (to download all results, please select display number as all.).
⑤ Arrange: users can sort the form in ascending or descending order based on a column.
⑥ Related link: hyperlink is underlined in blue, and users can click to go to the relevant page.
⑦ Display number: the number of results displayed on the current page.
⑧ Page: users can turn pages here.