Abstract
Cutaneous melanoma and basal cell carcinoma (BCC) represent the two primary malignancies driven by solar ultraviolet (UV) radiation. To map the spatial topology and distance dependence of their mutational signatures, we deployed new analytical bioinformatics workflow utilizing two convergent strategies: a data-driven read-level phasing framework to discriminate between collateral mutational events versus independent ones and a hypothesis-driven spatial permutational simulation model to capture density-dependent local deviations. A whole-genome analysis employing this framework unexpectedly reveals two fundamentally distinct lesion-processing landscapes, present in both BCC and melanoma. While independent mutations consistently reproduce canonical UV signatures (SBS7a-c) in both tumor types, collateral mutations tell a distinct and more varied narrative between BCC and melanoma. These collateral mutations are unusually abundant for non-UV characteristics, such as age-related SBS1 and SBS5, and display a notable 3' to 5' asymmetry near UV-induced lesions in pyrimidine dimers. We also demonstrated that BCC displays a pronounced enrichment of dinucleotide base substitutions flanking UV-signature sites on the 3' side, particularly CA>TG and CG>TA changes. Ultimately, these topological patterns in both tumors indicate that a primary photoproduct seeds secondary mutagenesis within its local chromatin environment, dramatically altering our understanding of UV-induced lesion processing.