Publication: Beyond Complementarity: A Coarse-Grained Simulation Pipeline for Antisense Oligonucleotides
Files
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Access Restrictions
Abstract
Antisense oligonucleotides (ASOs) are an established therapeutic modality with a growing preclinical pipeline. Translating a transcript into a viable ASO, however, requires more than sequence complementarity: mRNA folds into secondary and tertiary structures that render nominally complementary sites base-paired, buried, or only transiently accessible. Existing in silico tools capture secondary structure or single static three-dimensional folds but do not report time-dependent accessibility or binding behaviour in the presence of an ASO. This thesis develops a coarse-grained molecular dynamics (CGMD) pipeline that narrows these gaps. Target structures are obtained from the Protein Data Bank or predicted with NuFold and simulated in LAMMPS under the single-interaction-site (SIS) RNA force field of Nguyen, Hori, and Thirumalai [1] with the folding and binding implementation of Aierken and Joseph [2]. Candidate sites are identified by intersecting secondary-structure single-strandedness with per-residue solvent-accessible surface area, and ASO–RNA association is scored from the trajectory as a comparative apparent dissociation constant under a centre-of-mass distance framework. The pipeline is applied to a 1YMO-derived hTR pseudoknot hairpin and extended, without per-target retuning, to six additional PDB-derived RNAs. Across ten ASO variants at fixed loading, apparent Kd spans roughly an order of magnitude; extended-complementarity designs register the highest apparent Kd despite suppressing loop-core fluctuation most strongly, and both loop-core RMSF and apparent Kd depend non-monotonically on ASO loading. The integrated ranking reorders candidates relative to a complementarity-only shortlist—the pipeline’s principal contribution. Apparent Kd is reported as a comparative rather than an absolute quantity, several comparisons are single-replicate, and cross-target comparisons use differing binding criteria and box volumes; these caveats attach to the absolute values, not to the reordering itself.