Introduction

This page provides notes and instructions on how to explore, visualize and extract drug-similarity information in DrugSimDB.

Current version of DrugSimDB includes chemical, structure, pathway, and functional GO terms similarities of 10,317 small-molecule drugs, yielding 238,635 drug-pairs with significant aggregated score.

Overall, this interface is user-friendly and responsive.

Main Search Interface

The main interface for exploring the DrugSimDB database looks like bellow:

Search a Drug

The following image shows the search panel, where a drug name is queried in the DrugSimDB database to retrieve its similarity information with other drugs. Please note DrugSimDB only includes similarity information about statistically significant drug-pairs (total: 238,635).

It lists all the available drugs that have similarity information in the DrugSimDB, including a search bar and an option for choosing the combined similarity score to display with, i.e. aggregated score, p-value or adjusted p-value. The search bar supports auto-complete feature (see below). Therefore, when user enters a partial keyword as the drug name, for example, "biva", the list automatically filters to matching drug names "Bivalirudin" and "Tetrahydrocannabivarin".

Each Drug name has a "View" button next to it, and when clicked, the result is displayed in the right panel that includes all the similarity information about the query drug (see next sub-section).

Query Result (Tabular-View)

When a the View button is pressed for a drug, all other drugs that have statistically significant combined-similarity will be shown in a table along with other individual similarity scores, e.g. structure similarity, chemical similarity, target pathway similarity, GO terms similarity. The table also shows the statuses of the incident drugs annotated in Drugbank. Note, depending on the option chosen (see above) before the query being made, the final column in the table displays the information about the combined knowledge of all the individual similarity scores, i.e. Combined score, p-value, or adjusted p-value.

This table view is exportable in multiple format, e.g. CSV, Excel, PDF. It can also be copied in the ClipBoard and be printed using system default printer driver. There is a global search bar at the top of the view, which allows user to filter any the whole table based on a query word that works for all the columns. Morever, the table can be sorted based on individual column and has pagination support.

Query Result (Network-view)

One of the most important feature of this interface is the network view of drug-pairs based on the similarity. Including the query drug, a subset of all the other drugs that are reported as being significantly similar (see the Table-View) in the DrugSimDB, has been used to extract the induced sub-network from the DrugSimDB network (undirected, formed by all the drug-pairs). The network-view of that induced sub-network looks like the following:

This view has several useful feature, which are described below.

Interactive

This view is interactive. Users can drag the components of the network, i.e. nodes or edges using mouse or touch in order for better visualization. Moreover, the view-pan has navigation button to move the whole network-view left, right, up, or down. There are buttons for zooming-in, zooming-out and expanding the network to fit the pan. Usuers can also export the whole view-pan, which only supports .png format only.

Visualy informative

IN this network-view, the nodes (Drugs) are colored based on the groups that they were annotated in the Drugbank, e.g. approved; approved, experimental; and approved, vet_approved. All the edge colors are the same, but their widths varies in proportionate to the scores shown in the final column (combined score, or p-value, or adjusted p-value) of the tablur-view.

Query-able

This network-view is query-able.

• Node: On node mouse-hover event, user can see the following informaiton

  1. Drug Name
  2. Status, i.e. Approved, investigational, etc.
  3. List of adverse Side-effects (from SIDER database)

  

• Edge: On click-event of an edge, it makes a pubmed query (using easyPubMed R package) with the two incident drug names, and retrieves and p a list of pubmed articles that have the co-occurance of those two drugs. This process has several steps:

  • Step #1: Using both drug names (incident to an edge), Make a query in PubMed in all the fields, a count indicating total number of article that co-mentions the drug-pair, is return. For example, for an edge with drug-pair, Apixaban-Bivalirudin Apixaban[All Fields] AND Bivalirudin[All Fields]
  • Step #2:
  • Step #3:

Query drug info

TBA

Download

TBA

Statistics

TBA

Issues/Bugs

TBA