Data source: DDInter

[newgene]

name: Curated Drug-Drug Interactions Database
url: http://ddinter.scbdd.com/
download: http://ddinter.scbdd.com/download/
license: CC-BY-NC-SA (http://ddinter.scbdd.com/terms/)

[mnarayan1]

I can claim this ticket.

[andrewsu]

emailed data owners to ask about providing mappings between DDinterIDs and DrugBank/ChEMBL/PubChem.

[mnarayan1]

Here's a parser I wrote with the data that's available so far.

[mnarayan1]

I was able to scrape additional information (PubChem, ChEMBL, DrugBank ids) from http://ddinter.scbdd.com/.

Web scraper: https://github.com/mnarayan1/DDInter/blob/main/webscraper.py
Scraped drug information (used directly in the parser): https://github.com/mnarayan1/DDInter/blob/main/drug_data.json

Updated parser: https://github.com/mnarayan1/DDInter/blob/main/parser.py
Sample output file: https://github.com/mnarayan1/DDInter/blob/main/ddinter_data.json

Sample Record:

{
   "_id":"DDInter1263_DDInter1_Moderate",
   "drug_a":{
      "ddinterid_a":"DDInter1263",
      "name":"Naltrexone",
      "chembl":"CHEMBL19019",
      "pubchem":"5360515",
      "drugbank":"DB00704"
   },
   "drug_b":{
      "ddinterid_b":"DDInter1",
      "name":"Abacavir",
      "chembl":"CHEMBL1380",
      "pubchem":"441300",
      "drugbank":"DB01048"
   },
   "level":"Moderate"
}

[andrewsu]

@mnarayan1 in https://github.com/mnarayan1/DDInter/blob/main/webscraper.py#L7, you hard code a range on DDInter identifiers. Is it possible to generalize that so that your parser would work seamlessly when/if DDInter is updated? Thinking of iterating until you get a 404, or perhaps basing the scaping based on DDInter values in the downloadable file?

[mnarayan1]

I've updated the scraper to iterate until a page doesn't exist. Let me know if there's anything else I should fix.

[andrewsu]

Have you confirmed that all of the IDs run continuously with no interruption? For example, is there ever the case where DDInter1234 was retired or for some reason doesn't exist?

[mnarayan1]

So far, all of the IDs exist and run continuously.

[andrewsu]

great, I think this data plugin is ready for @erikyao to deploy. In parallel, you can also start working on writing the SmartAPI / OpenAPI annotation, as described in https://github.com/biothings/BioThings_Explorer_TRAPI/blob/main/docs/README-writing-x-bte.md. That guide is a work in progress, so suggestions on improvements are welcome and questions can go to Colleen and/or Rohan.

[erikyao]

Hi @andrewsu, I am testing the parser and I think the field names "ddinterid_a" and "ddinterid_b" could be better named "ddinterid", or more simply "ddinter". What do you think? Thanks!

[andrewsu]

Great point and good catch. Yes, let's go with ddinter. Thanks!

[erikyao]

Hi @mnarayan1, I found multiple documents with {'_id': 'DDInter424_DDInter7_Major'}. Please let me know when you fix the issue. Thanks!

P.S. I have forked your repo to https://github.com/biothings/DDInter and made some changes. Please let me know if you want to work directly on that fork (I can make a PR on your request).

[mnarayan1]

@erikyao I updated my parser, so hopefully the problem should be fixed. Could you make a pull request so I can update my repo with the changes you made? Thanks!

[erikyao]

@mnarayan1 Please find the PR at mnarayan1/DDInter#1. Thanks!

[andrewsu]

the PR above was merged and deployed. Next step is to write SmartAPI / x-bte annotation.

[andrewsu]

As noted in the linked issue just above:

The list of downloadable files at http://ddinter.scbdd.com/download/ appears to be incomplete. For example, the interaction noted by @karafecho above (http://ddinter.scbdd.com/ddinter/interact/1002248/) is not found in any of the linked files, but it is found in the unlinked http://ddinter.scbdd.com/static/media/download/ddinter_downloads_code_N.csv file

I think this parser should be updated to download and process all files corresponding to http://ddinter.scbdd.com/static/media/download/ddinter_downloads_code_[A-Z].csv

[mnarayan1]

I updated the parser to download all the files in http://ddinter.scbdd.com/static/media/download/ddinter_downloads_code_[A-Z].csv. Some letters don't seem to have a file (ex. there is no ddinter_downloads_code_E).

[colleenXu]

Checking with @mnarayan1 @andrewsu :

• It sounds like we'll want to update / re-deploy this API?
• I wonder if we can get more information, like:
  • the Interaction ID. We can then use it to construct linkouts to the database like http://ddinter.scbdd.com/ddinter/interact/1047495/ (either in the biothings API or in jq annotation for BTE)
  • the mechanism (aka what is the interaction caused by). See this page for an example of a drug's interactions that show many different mechanism categories.
  • description of the interaction, references, alternatives for either drug, possible metabolism interactions. These are sections of interesting data that can be seen on a webpage for an interaction: http://ddinter.scbdd.com/ddinter/interact/981427/

[tokebe]

This API has been deployed to Prod, should the issue be closed?

[colleenXu]

Yep, closing it now