Deciphering complex breakage-fusion-bridge genome rearrangements with Ambigram

Article https://doi.org/10.1038/s41467-023-41259-w Deciphering complex breakage-fusionbridge genome rearrangements with Ambigram Received: 13 October 2022 Check for updates 1234567890():,; 1234567890():,; Accepted: 28 August 2023 Chaohui Li 1,2, Lingxi Chen Shuai Cheng Li 1 1,2 , Guangze Pan1, Wenqian Zhang1 & Breakage-fusion-bridge (BFB) is a complex rearrangement that leads to tumor malignancy. Existing models for detecting BFBs rely on the ideal BFB hypothesis, ruling out the possibility of BFBs entangled with other structural variations, that is, complex BFBs. We propose an algorithm Ambigram to identify complex BFB and reconstruct the rearranged structure of the local genome during the cancer subclone evolution process. Ambigram handles data from short, linked, long, and single-cell sequences, and optical mapping technologies. Ambigram successfully deciphers the gold- or silver-standard complex BFBs against the state-of-the-art in multiple cancers. Ambigram dissects the intratumor heterogeneity of complex BFB events with single-cell reads from melanoma and gastric cancer. Furthermore, applying Ambigram to liver and cervical cancer data suggests that the BFB mechanism may mediate oncovirus integrations. BFB also exists in noncancer genomics. Investigating the complete human genome reference with Ambigram suggests that the BFB mechanism may be involved in two genome reorganizations of Homo Sapiens during evolution. Moreover, Ambigram discovers the signals of recurrent foldback inversions and complex BFBs in whole genome data from the 1000 genome project, and congenital heart diseases, respectively. Breakage-fusion-bridge (BFB) is a mechanism that leads to complex genome rearrangements in multiple cancers1–13. The rearrangement of BFB is mediated by the recursive cycles of BFB14–16 (Fig. 1a). A BFB cycle begins with the fold-back inversion (FBI) of two sister chromatids due to the lack of telomeres during DNA replication, resulting in twocentromeres in the fused bridge. When two centromeres are stretched to opposite poles in the anaphase, the two sister chromatids are split with a double-strand break on the bridge between two centromeres, unnecessarily the same as the previous fusion site. Since each daughter cell contains chromatids without telomeres, another BFB cycle may start again. Repetition of BFB cycles contributes to a surge in stair-like copy number (CN) amplifications and FBIs15–18. The above depicts a perfect BFB event that is solely driven by reverse complementary FBI, where the genomic positions of two breakpoints of a reverse complementary FBI are the same. However, some BFB events involve imperfect FBI whose breakpoint positions are different, resulting in the loss of DNA segments near the breakpoints4. Furthermore, studies reported that structure variations (SVs) such as deletion, duplication, insertion, and translocation could be involved during BFB cycles outside the FBI breakpoints13,19–21. In this study, we coin the BFB rearrangement beyond perfect BFB as complex BFB rearrangement. As the BFB process delivers anaphase bridges and dicentric chromosomes, investigators detected it using classical cytogenetic techniques in the early time10. However, these BFB cytogenetic signatures are not directly discernible from high-throughput DNA sequencing reads. Most studies inferred the consistency of a specific observation with BFB events from sequencing reads by two distinct hallmarks3–9,11,12,16: (i) oscillating CN with exponential or stair-like gains; 1 Department of Computer Science, City University of Hong Kong, Hong Kong, China. 2These authors contributed equally: Chaohui Li and Lingxi Chen. e-mail: Nature Communications | (2023)14:5528 1 Article https://doi.org/10.1038/s41467-023-41259-w a c Mechanism of BFB Telomere loss Locate reference path Fusion FBI Workflow of Ambigram Find BFB candidate Bridge Fusion m 1,2 = 1 2 Refine CN of BFB patterns and loops (ILP) ILP Results: Construct DAG (Directed Acyclic Graph) Accumulation of BFB recursive cycles b 8 Resolve the local genomic map (LGM) of BFB d Hallmarks of BFB 1,2 = m Bridge 3 2 Enumerate BFB mono-chains (m) and loops () Breakage FBI 1 1 2 4 3,5 = 1, 1,2 = 1,2 = 0, … 1 2 3 4 5 6 (3,6) 6 5 4 3 6 2 1, … 2 1 4,5 = 1, (4,6) 5 3 4 5 2 1 1 2, … 3,6 = 1, 1,6 = 1 (3,5) (4,5) 5 4 4 5 5 4 3 3 4 5 6| 6 5 4 3 2 1 Metrics of Ambigram 8 6 Stair-like CN gains 4 2 2 Fold-back inversions e Complex BFB (DEL/DUP/INS/TRX) Short/linked/ long reads, optical mapping Single-cell reads Onco-virus integration Ambigram resolves complex BFB events on pan-cancer and complex disease Melanoma Lung cancer Breast cancer Pancreatic cancer Gastric cancer Liver cancer Cervical cancer Heart disease Fig. 1 | Schematic overview. a The mechanism of BFB. b The hallmarks of BFB. c Workflow of Ambigram. d Metrics of Ambigram. e Summary of Ambigram applications. All anatomy images with free licenses are provided by Freepik. BFB breakage-fusion bridge, FBI fold-back inversion, CN copy number, ILP integer linear programming, DEL deletion, DUP duplication, INS insertion, TRX translocation. and ii) enrichment of FBIs with the fold-back direction of head-to-head (FBI-hh) or tail-to-tail (FBI-tt) (Fig. 1b). However, a pattern consistent with the two hallmarks does not imply that BFB yields the pattern. Leveraging the “palindrome” or “ambigram” nature of the BFBinduced local genomic map, i.e., the rearranged structure of the local genome, researchers started to utilize well-established algorithms to mathematically expand CNs or FBIs into possible BFB paths for array comparative genomic hybridization (aCGH) or pair-end sequencing (PE) data (Kinsella et al.22, BFBFinder23,24, and Greenman et al.25,26, Table 1). However, these mathematical models have limitations in interpreting real-world data for the following reasons. (i) These models solely focus on perfect BFB, ruling out the possibility of other deletion, duplication, insertion, or translocation in complex BFB rearrangements13,19–21. (ii) These models lag behind recent advances in linked read sequencing27,28, long-read sequencing29,30, and optical mapping alignment31. Linked read, long read, and optical mapping data offer larger than 100 kb linkage of the DNA fragment that could accelerate the detection of the accurate and long complex BFB local genomic map. (iii) Single-cell sequencing32–35 could facilitate investigation of the intratumor heterogeneity of complex BFB at single-cell resolution while existing methods are incompatible. Recently, the community has developed computational pipelines to detect and resolve complex somatic genome rearrangements, including complex BFB (AmpliconArchitect + AmpliconClassifier19, AmpliconReconstructor20, and LINX21, Table 1). However, these pipelines merely support short sequencing reads or optical mapping data. In this work, to overcome the limitations above, (...truncated)