NAISS
SUPR
NAISS Projects
SUPR
Scaling GPU-accelerated multiscale simulations of protein–DNA recognition and DNA-based biosensing systems
Dnr:

NAISS 2026/3-568

Type:

NAISS Medium

Principal Investigator:

Gioacchino Schifino

Affiliation:

Uppsala universitet

Start Date:

2026-06-24

End Date:

2027-07-01

Primary Classification:

10610: Bioinformatics and Computational Biology (Methods development to be 10203)

Secondary Classification:

10408: Biochemistry

Tertiary Classification:

10402: Physical Chemistry

Webpage:

Allocation

Abstract

This project will scale up an ongoing NAISS Small pilot effort into a production-level programme of GPU-accelerated molecular simulations of protein–DNA recognition, DNA topology, and DNA-based biosensing systems. The work is connected to computational chemistry, structural biophysics, and biosensor-oriented molecular design at the Department of Chemistry – BMC, Chemistry for Life Sciences, Uppsala University. The project will focus on three complementary directions. First, TopR1–DNA complexes and selected TopR1 variants will be simulated to investigate DNA recognition, DNA deformation, catalytic-site organization, and variant-dependent structural effects. Second, protein–DNA and aptamer-related systems will be modelled to study conformational dynamics, molecular recognition, and the stability of nucleic-acid/protein interfaces relevant to biosensing. Third, coarse-grained DNA simulations will be used to explore larger-scale DNA conformational behaviour and topology-dependent structural states that cannot be exhaustively sampled with atomistic simulations alone. The computational workflow will combine atomistic molecular dynamics, mainly with GROMACS, and coarse-grained DNA modelling using oxDNA. The current Small allocation is being used for setup, workflow validation, benchmarking, and initial test simulations. The present Medium request is needed to move from this pilot stage to systematic production simulations, including multiple independent replicas, alternative starting conformations, selected variants, and control systems. The requested Arrhenius GPU allocation is necessary because the systems are large, flexible, and require extensive sampling to distinguish robust molecular behaviour from trajectory-specific fluctuations. The workload is well suited to Arrhenius GPU because most simulations can be run as independent single-GPU jobs, enabling efficient throughput across many replicas and systems. The expected outcome is a molecular-level understanding of how protein–DNA recognition, DNA deformation, sequence or variant effects, and nucleic-acid conformational dynamics contribute to biosensor-oriented molecular design. The project will provide computational support for ongoing experimental and modelling activities in structural biophysics, DNA nanotechnology, and biomolecular sensing.