We want to identify clusters of triplet components with similar histories. These might arise due to e.g. a pulse in a mutational signature that affects multiple triplets, or paternal age effects.
Approaches
[x] perform independent $\mu$ inference for each of the 96 triplet spectrum components, and then post hoc clustering
[x] model the full 3FS jointly and assume something sparsity induing on the 3-\mu history (i.e. low rank)
We want to identify clusters of triplet components with similar histories. These might arise due to e.g. a pulse in a mutational signature that affects multiple triplets, or paternal age effects.
Approaches