Fetching the disease ontology level

Data Access GraphQL API
1

Hi kind people of the OT community!
I wanted to know if it is possible, using the API, to fetch the level of a disease in the disease ontology?
So for example, in the image - colitis is positioned 5th from the root. Is it possible to get this number using the API?

image
image1710Ă—298 32.9 KB

2

Hi Aner,

We don’t provide the level in the API. It is relatively complicated to get, as you can:

  • Have different levels of depth depending on which Therapeutic Area you use as root (diseases can have more than one).
  • Have multiple paths with different lengths from the disease to the Therapeutic Area .

In the next release (which is coming very soon, stay tuned!), we’ve changed the disease object to include a resolvedAncestors field which contains the list of all ancestors up to therapeutic areas as resolvable objects. With that you will be able to query like:

{
  disease(efoId: "MONDO_0011913") {
    id
    parents {
      id
    }
    resolvedAncestors {
      id
      parents {
        id
      }
    }
  }
}

And you would be able to get the level for every Therapeutic Area relevant to the disease by computing a DFS for each and storing the longest path. Here is an example in Python:

import requests

def get_node_levels(data):
    disease = data['data']['disease']
    roots = {n['id']: 1 for n in disease['resolvedAncestors'] if not n['parents']}
    nodes = {disease['id']: disease, **{n['id']: n for n in disease['resolvedAncestors']}}

    def max_path_to_root(current_id, root_id, visited=None):
        if visited is None:
            visited = set()
        if current_id == root_id:
            return 1
        if current_id in visited:
            return -1
        visited.add(current_id)
        paths = [max_path_to_root(p['id'], root_id, visited.copy()) for p in nodes[current_id]['parents']]
        valid_paths = [p + 1 for p in paths if p != -1]
        return max(valid_paths) if valid_paths else -1

    for root_id in roots:
        max_path = max_path_to_root(disease['id'], root_id)
        if max_path != -1:
            roots[root_id] = max_path
    return roots

query = lambda d: requests.post(
    "https://platform.opentargets.org/api/v4/graphql",
    json={
        "query": """
        {
            disease(efoId: "%s") {
                id
                parents {
                    id
                }
                resolvedAncestors {
                    id
                    parents {
                        id
                    }
                }
            }
        }
        """ % d
    }
).json()

print(get_node_levels(query("MONDO_0011913")))

❯ python levels.py
{'MONDO_0002025': 8, 'OTAR_0000018': 7, 'EFO_0000618': 10}

I’ve picked up a disease with a relatively complex graph, but there might be some that are even worse.

1 Like
3

Thank you. I think that the “resolvedAncestors” object would work great. Looking forward for that.