Nataša Miškov-Živanov

BELL automates evidence retrieval and scoring for biological interactions, achieving high confidence in 49.5% of tested interactions and guiding curators effectively.

This paper constructs a knowledge graph of CAR T cell signaling interactions using automated text mining and large language models to aid in CAR design.

NL2BLTL automates translating natural language biological hypotheses into bounded linear temporal logic specifications, achieving high accuracy and syntactic validity.

The paper introduces CELESTA, a framework for biomedical context classification that improves accuracy by integrating semi-supervised multi-task learning and a large-scale annotated dataset.

This work improves a GBM mechanistic model and evaluates NLP and LLM approaches for model refinement using graph metrics and literature curation.

The paper introduces the BioRECIPE format for representing biological models to enable human-machine interaction and model development.

VIOLIN reconciles heterogeneous interaction graphs with extracted molecular interactions, showing stable and interpretable results across different models and datasets.

ACCORDION selects relevant knowledge and recommends models by integrating literature retrieval, graph algorithms, and simulation, improving accuracy and reducing errors in biological systems.

This paper introduces SNAP-CAR and SNAP-synNotch receptors that can be post-translationally targeted by BG-conjugated antibodies to activate T cell signaling or gene programs in an antigen-specific manner.

The paper extracts and represents TMS-derived brain connectivity data in a graph database to explore interregional communication and temporal aspects of brain networks.

The paper integrates three tools to enhance automated model extension from literature, addressing issues like machine reading errors and high literature volume.

This paper develops automated methods to verify and assemble a GBM stem cell signaling network, showing that automated approaches lack precision and realistic topology compared to hybrid methods.

The paper develops a computational model of pancreatic cancer signaling and mutations to identify potential inhibitor combinations for therapy.