Testing the Virus Analogy for Malware: Simulation Checks of Clock and Tree-Adequacy Tests on Sandbox Behaviour
Malware variants arise by change and reuse, so they may form trees. The virus analogy also suggests a molecular clock. We test both with tools from evolutionary biology. We apply a root-to-tip clock test, a reticulation test and binary-character likelihood models to sandbox behaviour. We check the tests on simulated evolution with known history, mutation, clocks, cross-family recombination and sandbox noise. All 2784 simulated datasets in five studies are complete. In one study, sampling dates cluster by clade and have no date-divergence link. At a pilot-like date spread, the plain clock test, the clade-clustered test and the Mantel test reject in 0.60, 0.50 and 0.58 of datasets, far above the nominal level. Clustered sampling dates can therefore produce temporal signal that does not come from a link between dates and divergence. The reticulation test is above the nominal level in some cells. The worst Type I rate is 0.16, in a noisy cell. On two families from the Avast-CTU CAPEv2 reports, the unfiltered root-to-tip test is uninformative, because every tree has an edge at the branch-length bound. The Mantel test on collapsed Hamming distance rejects in both families. For unfiltered Emotet, the full-profile Hamming test does not reject. The delta statistic exceeds a tree-only null in every subsample, also in the one quality-controlled subsample without a capped edge. We read this as poor fit of a tree-like model, not as recombination. These pilot results are preliminary. We do not claim the first use of phylogenetic methods on malware. Preprint v0.1.0, not peer-reviewed. This record holds the paper (PDF) and a source archive of the code, stored results and paper sources at tag v0.1.0.
Authors
- Satwik Bhimavarapu (ORCID: https://orcid.org/0009-0002-4676-8851)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-12
- DOI
- https://doi.org/10.5281/zenodo.24159578
- Primary Topic
- Advanced Malware Detection Techniques
- Type
- preprint