Biomedical Text Mining and Ontologies

This paper develops an ontological framework for cellular genealogies to support integration of biological knowledge from single-cell experiments.

This paper reviews methods that integrate biomedical ontologies with machine learning to improve similarity computation and model performance using ontology embeddings and constraints.

This paper applies lexical link analysis to integrate unstructured CORD-19 literature with structured Next Strain genomic data to address high-priority questions about SARS-CoV-2.

The paper introduces MechSpy, a framework that generates mechanistic hypotheses of toxicity by combining gene expression data with biological knowledge, and validates predictions experimentally.

The paper introduces PheKnowLator, a framework for constructing large-scale biomedical knowledge graphs, and evaluates its performance on human disease mechanisms.

Lisen&Curate is a system that aids manual curation of bacterial transcription initiation regulation by integrating text mining and sentence similarity metrics to accelerate literature review and knowledge extraction.

The paper introduces L-Regulon, a soft-curation approach using semantic enrichment to improve knowledge extraction from scientific literature, specifically applied to RegulonDB data.

The paper introduces a method that uses deep Siamese neural networks to normalize biomedical ontologies and improve predictions of protein interactions and gene-disease associations.

The paper introduces a platform for real-time synthesis of biomedical knowledge using unsupervised neural networks, integrating unstructured literature with structured databases.

Pangaea extracts context-dependent gene relationships from biomedical literature using natural language processing and provides scalable, extensible tools for analysis.

The paper presents a system for automatic literature mining and summarization using deep learning to accelerate research during pandemics, specifically for COVID-19.

This paper introduces MeSHHeading2vec, a method that represents MeSH headings as vectors using graph embedding algorithms, showing improved performance in node classification and relationship prediction tasks.

The paper describes Wikidata as a FAIR knowledge graph for life sciences, showcasing its breadth, tools for integration, and use cases in biomedical analysis.

The paper develops an ontology and knowledge graph to model heme biology and investigates its role in Toll-like receptor signaling.

The paper develops a benchmark dataset and two machine learning models to classify compound-protein relationships in scientific abstracts as functional or non-functional.

DISNET extracts phenotypic disease information from public sources to create customizable disease networks for drug repositioning.

This paper introduces a post-processing algorithm that converts raw NLP output from EHRs into usable medication dose data for analysis.

This paper evaluates neural network models for embedding Gene Ontology terms and introduces new models based on GCN, ELMo, and BERT, finding that BERT-based embeddings perform best for rare labels.

The paper translates plant phenotype descriptions into structured formats using computational methods and shows they perform comparably to manual curation for identifying genetic associations.

BEE provides a natural language query interface for exploring relationships among biomedical entities across multiple datasets.

PGxCorpus provides a manually annotated dataset of 945 sentences to extract pharmacogenomics relationships from text, enabling supervised machine learning applications in the field.

Vapur is a search engine that finds protein-chemical pairs in COVID-19 literature using a relation-oriented inverted index.

The paper presents a structured review of NGLY1 Deficiency as a knowledge graph, improving hypothesis generation through FAIR principles and enabling collaborative curation.

ModEx extracts mode of regulation (activation/repression) for TF-gene interactions from ChIP-Seq data using text mining and achieves high F-scores on curated datasets.

The paper evaluates how formal axioms and ontology metadata improve ontology-based predictions of protein-protein interactions and gene-disease associations.

The paper presents a deep learning model for biomedical relation extraction between bacteria and biotopes, achieving higher F1 scores than existing methods.

The paper describes the development of CIViCmine, a text-mining tool that extracts clinically relevant cancer biomarkers from literature to aid curation into the CIViC database.

This paper develops a knowledge graph embedding method to identify drug repurposing opportunities for rare diseases by integrating biomedical literature and validating hypotheses through external sources.

GeDex extracts consensus gene-disease associations from biomedical literature using frequency patterns and supervised learning, achieving competitive performance compared to existing methods.

The paper updates UniProtKB enzyme annotations using Rhea, replacing textual descriptions with structured biochemical reaction data and improving search capabilities.