search

NCATSTranslator / Feedback

A repo for tracking gaps in Translator data and finding ways to fill them.
7 stars 0 forks source link

rank lower predicates patterns that might "treat" or "cause" #444

Open sandrine-m opened 1 year ago

sandrine-m commented 1 year ago

Issue connected to query described in issue #443 :

Type: Bug Report URL: https://ui.test.transltr.io/results?l=Maturity-onset%20Diabetes%20Of%20The%20Young&i=MONDO:0018911&t=0&q=325624e2-6053-4e2b-963c-9f13981f5070 ARS PK: 325624e2-6053-4e2b-963c-9f13981f5070 Steps to reproduce:

  1. MVP1 on Maturity-onset Diabetes Of The Young
  2. filter for "causes" string

A lot of top ranked results are causing the disease through mutations of genes: the predicates pattern is [causes increase/decrease] -> causes. As I understand the reasoning is that the gene-->causes--> diseases is not directional so it could potentially treat the disease or cause. Am I understanding properly here? The issue is that although those responses are particularly useful to build up new hypotheses, they might lead to the opposite of what I asked for and potentially red flags answers: Bisphenol A --> treats --> MODY

image

while there is a growing literature body on bisphenol A correlated with increasing risk.

sandrine-m commented 1 year ago

After meeting at the DM call today; those type of data gap/reasoning issues need more work and will be fixed post September