The structure of many dsDNA viral capsids is now known, in some cases from experiments dating back to the 1970s. Interestingly, even now not a single dsDNA viral genome structure has been solved experimentally. There are several reasons for this. The dsDNA genome is more mobile than the capsid, and so is at lower resolution. Aspects of the genome structure which do not have the same symmetry as the capsid may be averaged out. Tools for model building are less advanced for nucleic acids than they are for proteins. We have recently published (https://pub.epsilon.slu.se/33057/1/flores-s-c-et-al-20240228.pdf) the packaging scheme of two such viral genomes. However the method is interactive and thus has a learning curve. In this project we will automate the process of viral genome tracing using a Machine Learning method. The resulting software will be usable not only for viral genomes, but for any DNA having spiral or other regularity. A bit more storage might be nice, e.g. continuing NAISS 2024/23-469.