G-quadruplex structures are key modulators of somatic structural variants in cancers.
Rongxin ZhangHuiling ShuYuqi WangTiantong TaoJing TuCheng WangJean-Louis MergnyXiao SunPublished in: Cancer research (2023)
G-quadruplexes (G4s) are non-canonical secondary genome structures. Aberrant formation of G4s can impair genome integrity. Investigation of the relationship between G4s and somatic structural variants (SVs) in cancers could provide a better understanding of the role of G4 formation in cancer development and progression. In this study, we combined bioinformatic approaches and multi-omics data to investigate the connection between G4s and the somatic SVs. Somatic SV breakpoints were significantly enriched in G4 regions, regardless of SV subtypes. This enrichment was only observed in regions demonstrated to form G4s in cells ("active quadruplexes"), rather than in regions with a sequence compatible with G4 formation but without confirmed G4 formation ("potential quadruplexes"). Several genomic features impacted the connection between G4s and SVs, with the enrichment being notably strengthened at the boundary of topologically associated domains. Somatic breakpoints were also preferentially associated with G4 regions with earlier replication timing and open chromatin status. In cancer patients with homologous recombination repair defects, G4s and somatic breakpoints were substantially more strongly associated. Machine learning models were constructed that showed that G4 propensity is a potent feature for predicting the density of SV breakpoints. Altogether, these findings suggest that the G4 structures play a critical role in modulating the production of somatic SVs in cancers.
Keyphrases
- cell cycle arrest
- copy number
- genome wide
- machine learning
- dna damage
- high resolution
- dna methylation
- gene expression
- dna repair
- transcription factor
- squamous cell carcinoma
- risk assessment
- childhood cancer
- deep learning
- oxidative stress
- signaling pathway
- electronic health record
- endoplasmic reticulum stress
- induced apoptosis
- neural network
- data analysis
- squamous cell