CTKP Castleman treats edge not returned from KG2.10.1 #419

amykglen · 2024-11-08T18:41:21Z

noticed that, curiously, the below query (castleman) returns 1 edge when sent directly to the CTKP CI API, but no edges when sent directly to the KG2 CI API). odd, since KG2.10.1 includes CTKP... need to look into why this is happening. possible that it's simply due to the fact that CTKP has been updated since the KG2.10.1 build, or maybe it has something to do with the KG2pre ingest, or another part of the system?

{
  "edges": {
    "t_edge": {
      "object": "on",
      "predicates": [
        "biolink:treats"
      ],
      "subject": "sn"
    }
  },
  "nodes": {
    "on": {
      "categories": [
        "biolink:Disease"
      ],
      "ids": [
        "MONDO:0015564"
      ]
    },
    "sn": {
      "categories": [
        "biolink:ChemicalEntity"
      ]
    }
  }
}

The text was updated successfully, but these errors were encountered:

edeutsch · 2024-11-08T18:46:48Z

maybe because the predicate in RTX-KG2 got switched to biolink:treats_or_applied_or_studied_to_treat ?

amykglen · 2024-11-08T19:41:44Z

don't think so - when I raise the predicate to treats_or_applied_or_studied_to_treat, a few in_clinical_trials_for edges come back from CTKP, but that's it.

saramsey · 2024-11-13T17:28:01Z

OK, I've created a little test script for this:

curl -X POST -d @query.json \
       -H 'Content-Type: application/json' \
       -H 'accept: application/json' \
       https://multiomics.ci.transltr.io/ctkp/query | jq > results.json

which uses the query graph from above, in file query.json. I'll post the results below.

saramsey · 2024-11-13T17:28:49Z

Results here (removing the logs section for brevity):

{
  "message": {
    "knowledge_graph": {
      "edges": {
        "e66fbfac-0740-3919-adb1-ca8604e33666": {
          "attributes": [
            {
              "attribute_type_id": "biolink:supporting_study",
              "attributes": [
                {
                  "attribute_type_id": "clinical_trial_phase",
                  "value": 4,
                  "value_type_id": "biolink:ResearchPhaseEnum"
                },
                {
                  "attribute_type_id": "tested",
                  "value": "yes"
                },
                {
                  "attribute_type_id": "primary_purpose",
                  "value": "TREATMENT"
                },
                {
                  "attribute_type_id": "intervention_model",
                  "value": "SINGLE_GROUP"
                },
                {
                  "attribute_type_id": "clinical_trial_status",
                  "value": "WITHDRAWN"
                },
                {
                  "attribute_type_id": "start_date",
                  "value": "2008-12"
                },
                {
                  "attribute_type_id": "study_size",
                  "value": 0,
                  "value_type_id": "metatype:Integer"
                },
                {
                  "attribute_type_id": "enrollment_type",
                  "value": "ACTUAL"
                },
                {
                  "attribute_type_id": "age_range",
                  "value": "18 years or older"
                },
                {
                  "attribute_type_id": "child",
                  "value": false,
                  "value_type_id": "metatype:Boolean"
                },
                {
                  "attribute_type_id": "adult",
                  "value": true,
                  "value_type_id": "metatype:Boolean"
                },
                {
                  "attribute_type_id": "older_adult",
                  "value": true,
                  "value_type_id": "metatype:Boolean"
                },
                {
                  "attribute_type_id": "brief_title",
                  "value": "Valganciclovir to Treat HHV-8 Associated Multicentric Castleman's Disease",
                  "value_type_id": "metatype:String"
                }
              ],
              "value": "NCT00361933",
              "value_url": "https://clinicaltrials.gov/study/NCT00361933?tab=table"
            },
            {
              "attribute_type_id": "clinical_trial_tested_intervention",
              "value": "yes",
              "value_type_id": "metatype:String"
            },
            {
              "attribute_type_id": "biolink:clinical_approval_status",
              "value": "biolink:approved_for_condition",
              "value_type_id": "biolink:ClinicalApprovalStatusEnum"
            },
            {
              "attribute_source": "infores:multiomics-clinicaltrials",
              "attribute_type_id": "elevate_to_prediction",
              "value": false,
              "value_type_id": "metatype:Boolean"
            },
            {
              "attribute_source": "infores:multiomics-clinicaltrials",
              "attribute_type_id": "intervention_boxed_warning",
              "value": "2/2"
            },
            {
              "attribute_source": "infores:multiomics-clinicaltrials",
              "attribute_type_id": "biolink:knowledge_level",
              "value": "knowledge_assertion"
            },
            {
              "attribute_source": "infores:multiomics-clinicaltrials",
              "attribute_type_id": "biolink:agent_type",
              "value": "manual_agent"
            },
            {
              "attribute_type_id": "biolink:max_research_phase",
              "value": 4,
              "value_type_id": "biolink:ResearchPhaseEnum"
            }
          ],
          "object": "MONDO:0015564",
          "predicate": "biolink:treats",
          "sources": [
            {
              "resource_id": "infores:clinicaltrials",
              "resource_role": "supporting_data_source"
            },
            {
              "resource_id": "infores:aact",
              "resource_role": "supporting_data_source",
              "upstream_resource_ids": [
                "infores:clinicaltrials"
              ]
            },
            {
              "resource_id": "infores:multiomics-clinicaltrials",
              "resource_role": "primary_knowledge_source",
              "source_record_urls": [
                "https://db.systemsbiology.net/gestalt/cgi-pub/KGinfo.pl?id=e66fbfac-0740-3919-adb1-ca8604e33666"
              ],
              "upstream_resource_ids": [
                "infores:aact"
              ]
            }
          ],
          "subject": "UNII:GCU97FKN3R"
        }
      },
      "nodes": {
        "MONDO:0015564": {
          "attributes": [],
          "categories": [
            "biolink:Disease"
          ],
          "name": "Castleman disease"
        },
        "UNII:GCU97FKN3R": {
          "attributes": [],
          "categories": [
            "biolink:ChemicalEntity"
          ],
          "name": "VALGANCICLOVIR"
        }
      }
    },
    "query_graph": {
      "edges": {
        "t_edge": {
          "object": "on",
          "predicates": [
            "biolink:treats"
          ],
          "subject": "sn"
        }
      },
      "nodes": {
        "on": {
          "categories": [
            "biolink:Disease"
          ],
          "ids": [
            "MONDO:0015564"
          ]
        },
        "sn": {
          "categories": [
            "biolink:ChemicalEntity"
          ]
        }
      }
    },
    "results": [
      {
        "analyses": [
          {
            "edge_bindings": {
              "t_edge": [
                {
                  "attributes": [],
                  "id": "e66fbfac-0740-3919-adb1-ca8604e33666"
                }
              ]
            },
            "resource_id": "infores:multiomics-clinicaltrials"
          }
        ],
        "node_bindings": {
          "on": [
            {
              "attributes": [],
              "id": "MONDO:0015564"
            }
          ],
          "sn": [
            {
              "attributes": [],
              "id": "UNII:GCU97FKN3R"
            }
          ]
        },
        "resource_id": "infores:multiomics-clinicaltrials"
      }
    ]
  }
}

saramsey · 2024-11-13T17:37:50Z

I see one edge in those results, with subject UNII:GCU97FKN3R and object MONDO:0015564, and source infores:clinicaltrials.

saramsey · 2024-11-13T17:42:51Z

I created another little test script to run the same query graph through the RTX-KG2 KP:

curl -X POST -d @query.json \
       -H 'Content-Type: application/json' \
       -H 'accept: application/json' \
       https://kg2cploverdb.ci.transltr.io/query | jq > results2.json

saramsey · 2024-11-13T17:43:25Z

Confirmed, no results:

sramsey-laptop:issue-2414 sramsey$ cat results2.json
{
  "logs": [
    {
      "level": "INFO",
      "message": "kg2c: Converting qnode on's 'ids' to equivalent ids we recognize",
      "timestamp": "Wed, 13 Nov 2024 17:42:03 GMT"
    },
    {
      "level": "INFO",
      "message": "kg2c: After expansion to descendant concepts, object qnode has 5 ids",
      "timestamp": "Wed, 13 Nov 2024 17:42:03 GMT"
    },
    {
      "level": "INFO",
      "message": "kg2c: Looking up answers to query..",
      "timestamp": "Wed, 13 Nov 2024 17:42:03 GMT"
    },
    {
      "level": "INFO",
      "message": "kg2c: Found 0 input node answers, 0 output node answers, 0 edges",
      "timestamp": "Wed, 13 Nov 2024 17:42:03 GMT"
    },
    {
      "level": "INFO",
      "message": "kg2c: Beginning to transform answers to TRAPI format..",
      "timestamp": "Wed, 13 Nov 2024 17:42:03 GMT"
    },
    {
      "level": "INFO",
      "message": "kg2c: Done with query, returning TRAPI response (0 results)",
      "timestamp": "Wed, 13 Nov 2024 17:42:03 GMT"
    }
  ],
  "message": {
    "knowledge_graph": {
      "edges": {},
      "nodes": {}
    },
    "query_graph": {
      "edges": {
        "t_edge": {
          "object": "on",
          "predicates": [
            "biolink:treats"
          ],
          "subject": "sn"
        }
      },
      "nodes": {
        "on": {
          "categories": [
            "biolink:Disease"
          ],
          "ids": [
            "MONDO:0015564"
          ]
        },
        "sn": {
          "categories": [
            "biolink:ChemicalEntity"
          ]
        }
      }
    },
    "results": []
  }
}

saramsey · 2024-11-13T17:46:27Z

Looks like UNII:GCU97FKN3R corresponds to the antiviral "valganciclovir",
https://arax.ncats.io/?r=314336

saramsey · 2024-11-13T17:48:53Z

When I run the originally posted QG through arax.ci.transltr.io, I see "valganciclovir" as the third result,

and it is indeed provided by the Clinical Trials KP,

saramsey · 2024-11-13T17:49:38Z

Looks like ARAX on CI thinks that the preferred CURIE for valganciclovir is CHEBI:63635.

saramsey · 2024-11-13T17:57:43Z

So, there is a "treats" like edge between "valganciclovir" and "Castleman" in RTX-KG2.10.1c, see this Cypher query against the kg2canonicalized2.rtx.ai Neo4j endpoint:

Here is the Cypher query embedded as text:

match (n {id: 'CHEBI:63635'})-[r]->(m {id: 'MONDO:0015564'}) return n,m,r;

but note, the predicate is biolink:treats_or_applied_or_studied_to_treat, not biolink:treats.

saramsey · 2024-11-13T18:14:11Z

@amykglen points out that the above edge is actually from SemMedDB, not CTKP. Thank you, Amy! My bad!

amykglen · 2024-11-13T18:40:44Z

FYI - it looks like KG2c only contains CTKP's true predicates (no remapping, as desired):

MATCH (n)-[e]->(m) where e.primary_knowledge_source="infores:biothings-multiomics-clinicaltrials" return distinct e.predicate, count(distinct e)

e.predicate	count(distinct e)
"biolink:biolink_in_clinical_trials_for"	13459
"biolink:biolink_mentioned_in_trials_for"	14215
"biolink:biolink_treats"	3558

saramsey · 2024-11-13T19:03:45Z

Looks like this is a KG2pre bug. I ran the following on kg2101build.rtx.ai:

zcat kg2-simplified-2.10.1-edges.jsonl.gz | grep MONDO:0015564 | grep UNII:GCU97FKN3R | grep treats
ubuntu@ip-172-31-10-56:~/kg2-build$

saramsey · 2024-11-13T19:04:09Z

But we are clearly getting this edge in the CTKP data file dump:

ubuntu@ip-172-31-10-56:~/kg2-build$ grep MONDO:0015564 clinicaltrialskg_tsv_to_kg_jsonl-edges.jsonl | grep UNII:GCU97FKN3R | grep treats
{"domain_range_exclusion": false, "id": "UNII:GCU97FKN3R---biolink:biolink_treats---None---None---None---MONDO:0015564---ClinicalTrialsKG:", "negated": false, "object": "MONDO:0015564", "predicate": null, "primary_knowledge_source": "ClinicalTrialsKG:", "publications": [], "publications_info": {}, "qualified_object_aspect": null, "qualified_object_direction": null, "qualified_predicate": null, "relation_label": "biolink:treats", "source_predicate": "biolink:biolink_treats", "subject": "UNII:GCU97FKN3R", "update_date": "2008-12-01"}

saramsey · 2024-11-13T19:04:38Z

Transferring this issue to the RTX-KG2 project area

saramsey · 2024-11-13T19:35:58Z

So the edge does appear in clinicaltrialskg_tsv_to_jsonl-edges.jsonl file:

ubuntu@ip-172-31-10-56:~/kg2-build$ grep MONDO:0015564 clinicaltrialskg_tsv_to_kg_jsonl-edges.jsonl  | grep treats
{"domain_range_exclusion": false, "id": "UNII:GCU97FKN3R---biolink:biolink_treats---None---None---None---MONDO:0015564---ClinicalTrialsKG:", "negated": false, "object": "MONDO:0015564", "predicate": null, "primary_knowledge_source": "ClinicalTrialsKG:", "publications": [], "publications_info": {}, "qualified_object_aspect": null, "qualified_object_direction": null, "qualified_predicate": null, "relation_label": "biolink:treats", "source_predicate": "biolink:biolink_treats", "subject": "UNII:GCU97FKN3R", "update_date": "2008-12-01"}

saramsey · 2024-11-19T21:57:28Z

OK, the commits in issue #420 seem to fix this:

ubuntu@ip-172-31-13-50:~/kg2-build$ grep UNII:GCU97FKN3R kg2-merged-2.10.2-edges.jsonl | grep MONDO:0015564 | jq
{
  "domain_range_exclusion": false,
  "id": "UNII:GCU97FKN3R---biolink:in_clinical_trials_for---None---None---None---MONDO:0015564---ClinicalTrialsKG:",
  "negated": false,
  "object": "MONDO:0015564",
  "predicate": null,
  "primary_knowledge_source": "ClinicalTrialsKG:",
  "publications": [],
  "publications_info": {},
  "qualified_object_aspect": null,
  "qualified_object_direction": null,
  "qualified_predicate": null,
  "relation_label": "in_clinical_trials_for",
  "source_predicate": "biolink:in_clinical_trials_for",
  "subject": "UNII:GCU97FKN3R",
  "update_date": "2008-12-01"
}
{
  "domain_range_exclusion": false,
  "id": "UNII:GCU97FKN3R---biolink:treats---None---None---None---MONDO:0015564---ClinicalTrialsKG:",
  "negated": false,
  "object": "MONDO:0015564",
  "predicate": null,
  "primary_knowledge_source": "ClinicalTrialsKG:",
  "publications": [],
  "publications_info": {},
  "qualified_object_aspect": null,
  "qualified_object_direction": null,
  "qualified_predicate": null,
  "relation_label": "treats",
  "source_predicate": "biolink:treats",
  "subject": "UNII:GCU97FKN3R",
  "update_date": "2008-12-01"
}
{
  "domain_range_exclusion": false,
  "id": "UNII:GCU97FKN3R---biolink:mentioned_in_trials_for---None---None---None---MONDO:0015564---ClinicalTrialsKG:",
  "negated": false,
  "object": "MONDO:0015564",
  "predicate": null,
  "primary_knowledge_source": "ClinicalTrialsKG:",
  "publications": [],
  "publications_info": {},
  "qualified_object_aspect": null,
  "qualified_object_direction": null,
  "qualified_predicate": null,
  "relation_label": "mentioned_in_trials_for",
  "source_predicate": "biolink:mentioned_in_trials_for",
  "subject": "UNII:GCU97FKN3R",
  "update_date": "2011-09-13"
}

saramsey transferred this issue from RTXteam/RTX Nov 13, 2024

saramsey mentioned this issue Nov 13, 2024

RTX-KG2 needs to ETL the UNII identifiers list #420

Open

saramsey added the bug Something isn't working label Nov 13, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

CTKP Castleman treats edge not returned from KG2.10.1 #419

CTKP Castleman treats edge not returned from KG2.10.1 #419

amykglen commented Nov 8, 2024

edeutsch commented Nov 8, 2024

amykglen commented Nov 8, 2024 •

edited

Loading

saramsey commented Nov 13, 2024 •

edited

Loading

saramsey commented Nov 13, 2024 •

edited

Loading

saramsey commented Nov 13, 2024

saramsey commented Nov 13, 2024

saramsey commented Nov 13, 2024

saramsey commented Nov 13, 2024 •

edited

Loading

saramsey commented Nov 13, 2024

saramsey commented Nov 13, 2024

saramsey commented Nov 13, 2024 •

edited

Loading

saramsey commented Nov 13, 2024 •

edited

Loading

amykglen commented Nov 13, 2024

saramsey commented Nov 13, 2024

saramsey commented Nov 13, 2024

saramsey commented Nov 13, 2024 •

edited

Loading

saramsey commented Nov 13, 2024

saramsey commented Nov 19, 2024

CTKP Castleman treats edge not returned from KG2.10.1 #419

CTKP Castleman treats edge not returned from KG2.10.1 #419

Comments

amykglen commented Nov 8, 2024

edeutsch commented Nov 8, 2024

amykglen commented Nov 8, 2024 • edited Loading

saramsey commented Nov 13, 2024 • edited Loading

saramsey commented Nov 13, 2024 • edited Loading

saramsey commented Nov 13, 2024

saramsey commented Nov 13, 2024

saramsey commented Nov 13, 2024

saramsey commented Nov 13, 2024 • edited Loading

saramsey commented Nov 13, 2024

saramsey commented Nov 13, 2024

saramsey commented Nov 13, 2024 • edited Loading

saramsey commented Nov 13, 2024 • edited Loading

amykglen commented Nov 13, 2024

saramsey commented Nov 13, 2024

saramsey commented Nov 13, 2024

saramsey commented Nov 13, 2024 • edited Loading

saramsey commented Nov 13, 2024

saramsey commented Nov 19, 2024

amykglen commented Nov 8, 2024 •

edited

Loading

saramsey commented Nov 13, 2024 •

edited

Loading

saramsey commented Nov 13, 2024 •

edited

Loading

saramsey commented Nov 13, 2024 •

edited

Loading

saramsey commented Nov 13, 2024 •

edited

Loading

saramsey commented Nov 13, 2024 •

edited

Loading

saramsey commented Nov 13, 2024 •

edited

Loading