Context-driven interaction retrieval and classification for modeling, curation, and reuse
Luo, H.; Hansen, C.; Telmer, C. A.; Tang, D.; Arazkhani, N.; Zhou, G.; Spirtes, P.; Miskov-Zivanov, N.
Show abstract
Computational modeling seeks to construct and simulate intracellular signaling networks to understand health and disease. The scientific literature contains descriptions of experimental results that can be interpreted by machines using NLP or LLMs to itemize molecular interactions. This machine readable output can then be used to assess, update or improve existing biological models if there is a tool for comparing the existing model with the information extracted from the papers. Here we describe VIOLIN a tool for classifying machine outputs of molecular interactions with respect to a biological model. VIOLIN classifies interactions as corroborations, contradictions, flagged or extensions with subcategories of each class. This paper analyzes 2 different models, 9 reading sets, 2 NLP and 2 LLM tools to test VIOLINs capabilities. The results show that VIOLIN successfully classifies interaction types and can be combined with automated filtering to provide a versatile tool for use by the systems biology community.
Matching journals
The top 7 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Repurposing Non-pharmacological Interventions for Alzheimer’s Diseases through Link Prediction on Biomedical Literature 93%
- CONSORT-TM: Text classification models for assessing the completeness of randomized controlled trial publications 93%
- Embeddings from deep learning transfer GO annotations beyond homology 93%
Similar papers in this journal
- Benchmarking transformer-based models for medical record deidentification: A single centre, multi-specialty evaluation 93%
- Interactive Multiresolution Visualization of Cellular Network Processes 92%
- BATMAN: fast and accurate integration of single-cell RNA-Seq datasets via minimum-weight matching 92%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.