Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Merged ChemProt-DrugProt for Relation Extraction from Biomedical Literature

Mai H. Nguyen, Shibani Likhite, Jiawei Tang, Darshini Mahendran, Bridget T. McInnes

TL;DR

This work addresses the challenge of extracting chemical-gene relations from biomedical literature by creating a larger, unified benchmark through merging the ChemProt and DrugProt datasets. It evaluates two approaches, BioBERT for local sentence context and GCN-BERT that adds global document-context through a PMI-based vocabulary graph, demonstrating improved performance on CPR groups shared between datasets. Key contributions include a manual conflict-resolution merging workflow, training data augmentation with explicit negative samples, and systematic analysis of entity masking, downsampling, and weighted loss. The results show that the merged dataset enhances precision and recall for several CPR groups, and the combination of local and global context offers practical gains for biomedical relation extraction at scale.

Abstract

The extraction of chemical-gene relations plays a pivotal role in understanding the intricate interactions between chemical compounds and genes, with significant implications for drug discovery, disease understanding, and biomedical research. This paper presents a data set created by merging the ChemProt and DrugProt datasets to augment sample counts and improve model accuracy. We evaluate the merged dataset using two state of the art relationship extraction algorithms: Bidirectional Encoder Representations from Transformers (BERT) specifically BioBERT, and Graph Convolutional Networks (GCNs) combined with BioBERT. While BioBERT excels at capturing local contexts, it may benefit from incorporating global information essential for understanding chemical-gene interactions. This can be achieved by integrating GCNs with BioBERT to harness both global and local context. Our results show that by integrating the ChemProt and DrugProt datasets, we demonstrated significant improvements in model performance, particularly in CPR groups shared between the datasets. Incorporating the global context using GCN can help increase the overall precision and recall in some of the CPR groups over using just BioBERT.

Merged ChemProt-DrugProt for Relation Extraction from Biomedical Literature

TL;DR

This work addresses the challenge of extracting chemical-gene relations from biomedical literature by creating a larger, unified benchmark through merging the ChemProt and DrugProt datasets. It evaluates two approaches, BioBERT for local sentence context and GCN-BERT that adds global document-context through a PMI-based vocabulary graph, demonstrating improved performance on CPR groups shared between datasets. Key contributions include a manual conflict-resolution merging workflow, training data augmentation with explicit negative samples, and systematic analysis of entity masking, downsampling, and weighted loss. The results show that the merged dataset enhances precision and recall for several CPR groups, and the combination of local and global context offers practical gains for biomedical relation extraction at scale.

Abstract

The extraction of chemical-gene relations plays a pivotal role in understanding the intricate interactions between chemical compounds and genes, with significant implications for drug discovery, disease understanding, and biomedical research. This paper presents a data set created by merging the ChemProt and DrugProt datasets to augment sample counts and improve model accuracy. We evaluate the merged dataset using two state of the art relationship extraction algorithms: Bidirectional Encoder Representations from Transformers (BERT) specifically BioBERT, and Graph Convolutional Networks (GCNs) combined with BioBERT. While BioBERT excels at capturing local contexts, it may benefit from incorporating global information essential for understanding chemical-gene interactions. This can be achieved by integrating GCNs with BioBERT to harness both global and local context. Our results show that by integrating the ChemProt and DrugProt datasets, we demonstrated significant improvements in model performance, particularly in CPR groups shared between the datasets. Incorporating the global context using GCN can help increase the overall precision and recall in some of the CPR groups over using just BioBERT.
Paper Structure (29 sections, 9 equations, 3 figures, 6 tables)

This paper contains 29 sections, 9 equations, 3 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Related work
  3. Data
  4. Datasets
  5. Dataset Merging Processing
  6. Training Data
  7. Relation Extraction Methods
  8. BioBERT Model Setup
  9. GCN-BERT Model Setup
  10. Vocabulary Graph
  11. Graph Convolutional Network
  12. Combined GCN-BERT
  13. Experimental Setup
  14. Entity Masking
  15. Downsampling Negative Samples
  16. ...and 14 more sections

Figures (3)

  • Figure 1: Illustration of entities and relations in from sample input sentences
  • Figure 2: Architecture of GCN-BERT adapted from mahendran2022graph
  • Figure :