Oligonucleotide therapeutics offer groundbreaking possibilities for pharmacological intervention by enabling potent, sequence-specific modulation of principally any disease-related gene product. Thereby, diseases that are not possible to treat using traditional small-molecule drugs can now be addressed. The predominant subclass among current therapeutic oligonucleotides are the antisense oligonucleotides (ASOs). These drugs bind to their cognate mRNA transcripts through Watson-Crick base pairing, with the desired effect typically being degradation of the mRNA and a consequent knock-down of target protein levels. Despite the apparent simplicity of such oligonucleotide-target mRNA interactions, the molecular mechanisms of hybridization initiation and the dynamics of the binding process are poorly characterized. Furthermore, the vast number of potential off-target sites in a cell's transcriptome--i.e., sites which exhibit partial sequence matches to the antisense oligonucleotide--and the conformational dynamics of both the antisense oligonucleotide and the mRNA transcript, severely hampers predictions of optimal antisense sequences. Furthermore, therapeutic antisense oligonucleotides contain extensive chemical modifications, which improve stability to enzymatic cleavage, but which alters binding affinity relative to natural RNA/DNA oligonucleotides in unpredictable ways. Therefore, development programs for new antisense oligonucleotides typically utilize a costly, wet-lab trial-and-error process. To address these current knowledge gaps, this project therefore aims to 1) optimize and validate force field parameters for atomistic simulations of oligonucleotides, to include support for the common chemical modifications used in therapeutic antisense oligonucleotides, and use these to; 2) delineate the conformational dynamics of prototypical oligonucleotides and their intramolecular hybridization potential and; 3) decipher the molecular mechanisms of the target sequence recognition and duplex stability in antisense oligonucleotide--mRNA hybridization. In the previous access round, we developed force field parameters for a widely used chemical modification in therapeutic oligonucleotides, the phosphorothioate group, and used it to study conformational dynamics and duplex formation with target RNA. We will now expand to other chemical modifications, and evaluate oligonucleotide--RNA duplex dynamics using a combined MD-machine learning approach. Preliminary simulations indicate that all goals in the project are feasible, and previous experience suggests Dardel is a most suitable environment for the proposed activities.