Extended Data Fig. 1: Modification of Shatterseek by removing oscillating-copy-number-state requirement.
From: Starfish infers signatures of complex genomic rearrangements across human cancers
a, Examples of CGRs not detected by the original version of ShatterSeek. The SV and copy number profiles are shown for four CGRs. CGR regions are marked by red bars below SVs. b, Comparisons between CGR seed regions detected with and without oscillating-copy-state requirement. Breakpoint enrichment test is a binomial test corrected for mappability to evaluate the enrichment of SVs in each chromosome. Exponential distribution test evaluates whether the distribution of SV breakpoints differ from an exponential distribution. The smaller P values for breakpoint enrichment test and exponential test the better. Fragment joins test evaluates whether the distribution of DNA fragment joins diverges from a multinomial distribution with equal probabilities for each category using the goodness-of-fit test for the multinomial distribution. The larger P values for fragment joins test the better. FDR correction was performed on all P values. The newly detected CGRs without oscillating-copy-state have better P values for exponential distribution test and comparable P values for fragment joins test compared to CGRs detected by the original Shatterseek. CGRs detected with oscillating-copy-state have better P values in breakpoint enrichment test because more CGRs of Signatures 1 and 2 are detected with oscillating-copy-state (Figure S4c) and these CGRs have more SVs (Fig. 2b). Although the newly detected CGRs are less enriched in each chromosome, they all pass the Shatterseek P value cutoff.
