Infer hidden relationships from literature by multi level context terms
?
0 likes?217 views
This document proposes a method to infer hidden relationships between biological entities like genes and diseases by analyzing literature. It describes constructing context term vectors from literature about Alzheimer's disease to identify drug-disease interactions. These inferred interactions are then evaluated by comparing to known interactions from an existing ABC model, with the goal of improving on the ABC model's ability to find interactions requiring manual input from experts.
Infer hidden relationships from literature by multi level context terms
- 1. Infer Hidden Relationships From Literature By Multi-Level Context Terms
- 2. 1. Why? 2. How? 3. Evaluate Result Table of Contents
- 3. Why
- 4. ABC Model
- 5. ?Has many candidates ?Is semi-automatic ?Requires expert¡¯s manual input. ABC Model¡¯s Disadvantage
- 6. How Let¡¯s infer drugs that interact with Alzheimer¡¯s disease
- 7. 1. Construct Biological Data Entities 2. Collect Literatures about Alzheimer¡¯s disease 3. Extract Interactions with Context Term Vector 4. Infer Undiscovered Interactions
- 8. Biological Data Entities Gene, Drug, Disease, Symptom, Protein, Molecule, Process, Disease We need databases of
- 9. Download Literatures from PubMed
- 10. How to Extract Interaction 1. Make A-B and B-C context vector 2. Calculate A-B-C similarity score
- 12. Caculate Similarity Score ?Cosine similarity ?Spearman Correlation
- 13. Infer Undiscovered Interactions ?Sum of all A-B-C scores ?Maximum of A-B-C scores ?Count of some A-B-C scores ( > threshold)
- 14. Evaluate Result by Comparing with ABC model
- 15. 1. Top 100, 500, 1000 Interactions 2.Top 10 Interactions
- 16. Precision for ABC Model VS Similarity