Complex genomic rearrangements (CGRs) in cancer often originate from abnormal cellular structures such as micronuclei and chromatin bridges. However, the primary mechanisms responsible for CGR formation in disease tissues remain unclear, particularly due to the challenges in fully capturing these processes. To address this, we have developed “Starfish,” a computational algorithm to decipher CGR signatures and infer their forming mechanisms by analyzing distinctive copy number variations and breakpoint patterns. Here, we provide practical guidance on the application of “Starfish,” available as an R package, to study CGR signatures in human cancers.