Integrate UMLS Disease Mapping info into MyDisease.info

[kevinxin90]

Step 1: Extract all UMLS IDs belong to Disease Category

• source file: UMLS MRSTY file
• file location:
• This file maps a UMLS ID to its semantic type
  • The first column of the file corresponds to the UMLS ID
  • The fourth column of the file corresponds to the semantic type
• Extract all UMLS IDs belonging to these types
  • Disease or Syndrome
  • Mental or Behavioral Dysfunction
  • Neoplastic Process

Step 2: Extract UMLS Mapping

• source file: MRCONSO.RRF
• This file maps a UMLS ID to corresponding identifiers in other sources
  • The first column represents UMLS ID
  • The 2nd column represents language, we only need to parse rows with language equal to "ENG"
  • The 3rd column represents term status, could be P (preferred) or S (non-preferred)
  • group the data based on the value of 3rd column
  • The 12th column represents data source, we only need to parse rows with data source belong to these: NCI, MSH, SNOMEDCT_US, ICD10CM, ICD10, ICD10AM, ICD9CM.
  • rename the value of 12th column:
    • NCI: nci
    • MSH: mesh
    • SNOMEDCT_US: snomed
    • ICD10CM: icd10cm
    • ICD10: icd10
    • ICD10AM: icd10am
    • ICD9CM: icd9cm
  • The 14th column represents the value.

Step3: Query mydisease.info for the primary id used in mydisease.info.

• use batch query feature supported in MyDisease.info
• query 1000 UMLS ids at a time
• query syntax
  requests.post('http://mydisease.info/v1/query', data={'q': 'C0006826, C0008780', 'scopes':'disgenet.xrefs.umls, mondo.xrefs.umls'}, params={'fields': '_id'})
  replace the value of 'q' with the 1000 UMLS ids separated with comma
• If the primary id "_id" is not found, use the UMLS id as the primary id (prefixed with UMLS), e.g. UMLS:C0006826.
• If more than one primary id "_id" is found, create multiple records for it.

Example Output:

{
  "_id": "MONDO:0004992",
  "umls": {
      "mesh": {
             "preferred": "D009369",
             "non-preferred": "D009369"
      },
      "snomed": {
            "preferred": ["269513004", "154433003", ...],
            "non-preferred": ["38807002", ...]
      },
      "icd10": {
             ...
      },
      "icd9cm": {
             ...
      }
}