VARL Logo

Publications

Peer-reviewed research, conference papers, and preprints produced by VARL's scientific team. All publications are listed with full citation data and DOI links.

Filter
2024
Drug Discovery·Journal Article·November 2024

Predictive Toxicology via Multi-Organ Digital Twin Simulation

Vacid T., Köksal H., Tanaka K., Chen L.

Toxicological Sciences

A multi-organ digital twin system predicts compound toxicity across hepatic, renal, and cardiac tissues simultaneously. The platform reduces preclinical toxicity screening timelines by 87% while maintaining concordance with in vivo results at 93.4%.

DOI →
Bioinformatics·Journal Article·August 2024

Graph Neural Networks for Protein Interaction Prediction at Proteome Scale

Köksal H., Vacid T., Petrov D.

Bioinformatics

A graph neural network architecture that predicts protein-protein interactions across entire proteomes with 96.1% AUROC. The model operates on structural and sequence features simultaneously, enabling interaction prediction for proteins with no known experimental data.

DOI →
2025
Drug Discovery·Journal Article·December 2025

AI-Driven Target Identification in Autoimmune Pathway Networks

Köksal H., Vacid T., Chen L., Petrov D.

Cell Systems

A deep learning framework that identifies novel therapeutic targets within autoimmune signaling cascades by analyzing 4.2 million protein-protein interactions. The model achieves 94.7% validation rate against experimental assays and identifies 23 previously unreported druggable targets.

DOI →
Genomics·Journal Article·November 2025

Single-Cell Transcriptomic Atlas of Immune Response Dynamics

Arıkan M., Vacid T., Halvorsen I., et al.

Science

Comprehensive single-cell RNA sequencing atlas capturing immune response trajectories across 847 patients with varying autoimmune conditions. The dataset reveals previously uncharacterized cell state transitions that precede clinical disease onset by an average of 14 months.

DOI →
Agricultural Science·Journal Article·October 2025

Molecular Simulation of Crop Stress Response Under Climate Variability

Vacid T., Halvorsen I., Okonkwo M.

Nature Food

Digital twin models of wheat and rice immune systems predict stress responses under 47 distinct climate scenarios with 91.3% accuracy against field trial data. The framework enables pre-emptive breeding strategy optimization before environmental conditions materialize.

DOI →
Bioinformatics·Conference Paper·September 2025

varl-bench: A Standardized Benchmark Suite for Biological Prediction Models

Köksal H., Chen L., Vacid T.

Proceedings of ISMB 2025

We introduce varl-bench, an open benchmark suite comprising 14 standardized evaluation tasks for biological prediction models. The suite covers target identification, pathway analysis, biomarker detection, and treatment outcome prediction, establishing reproducible baselines across methodologies.

DOI →
Computational Biology·Journal Article·August 2025

Real-Time Biomarker Detection via Continuous Molecular Monitoring

Arıkan M., Tanaka K., Vacid T., et al.

Nature Biotechnology

A biosensor-integrated AI system that detects disease-associated biomarkers in real time from continuous blood monitoring data. The system identifies early-stage oncological markers with 97.2% sensitivity and 99.1% specificity across a validation cohort of 3,400 patients.

DOI →
Drug Discovery·Journal Article·June 2025

Generative Molecular Design for Targeted Protein Degradation

Vacid T., Köksal H., Petrov D., Vasquez E.

Journal of Medicinal Chemistry

A generative AI model that designs PROTAC molecules for targeted protein degradation with predicted binding affinity accuracy of 0.93 R² against experimental IC50 values. The model generated 12 novel degrader candidates, 8 of which showed activity in cellular assays.

DOI →
Genomics·Preprint·April 2025

Cross-Species Pathway Conservation and Divergence in Immune Regulation

Halvorsen I., Vacid T., Arıkan M.

bioRxiv

Comparative analysis of immune regulatory pathways across 7 model organisms reveals conserved core modules and species-specific divergence points. The findings enable cross-species translation of therapeutic targets with quantified confidence intervals for each translational step.

DOI →
Computational Biology·Journal Article·February 2025

Stochastic Noise Modeling in Biological Digital Twins

Köksal H., Vacid T.

PLOS Computational Biology

We propose a stochastic noise layer for biological digital twins that incorporates molecular-level variability into deterministic simulation frameworks. The model reproduces experimentally observed phenotypic heterogeneity in cell populations with 89% fidelity.

DOI →
Agricultural Science·Journal Article·January 2025

AI-Optimized Soil Microbiome Restoration Protocols

Vacid T., Okonkwo M., Halvorsen I.

Nature Sustainability

Machine learning models trained on 2,300 soil microbiome profiles generate restoration protocols that recover degraded agricultural soil biodiversity to 78% of reference ecosystem levels within a single growing season, validated across 14 field sites on 4 continents.

DOI →
2026
Computational Biology·Journal Article·January 2026

Digital Twin Architectures for Multi-Scale Biological Simulation

Vacid T., Köksal H., Arıkan M., et al.

Nature Computational Science

We present a novel architecture for constructing and operating digital twins of biological systems across molecular, cellular, and tissue scales. The framework enables real-time simulation of disease mechanisms with clinical-grade fidelity, validated against longitudinal patient cohort data spanning 12,000 subjects.

DOI →
ALSO FROM VARL

White Papers

In-depth technical reports, methodology documents, and research frameworks.

→