Chromatin interactions play important roles in regulating gene expression. However, the availability of genome-wide chromatin interaction data is limited. We develop a computational method, chromatin interaction neural network (ChINN), to predict chromatin interactions between open chromatin regions using only DNA sequences. ChINN predicts CTCF- and RNA polymerase II-associated and Hi-C chromatin interactions. ChINN shows good across-sample performances and captures various sequence features for chromatin interaction prediction. We apply ChINN to 6 chronic lymphocytic leukemia (CLL) patient samples and a published cohort of 84 CLL open chromatin samples. Our results demonstrate extensive heterogeneity in chromatin interactions among CLL patient samples.

Nasopharyngeal cancer (NPC), a cancer derived from epithelial cells in the nasopharynx, is a cancer common in China, Southeast Asia, and Africa. The three-dimensional (3D) genome organization of nasopharyngeal cancer is poorly understood. A major challenge in understanding the 3D genome organization of cancer samples is the lack of a method for the characterization of chromatin interactions in solid cancer needle biopsy samples. Here, we developed Biop-C, a modified in situ Hi-C method using solid cancer needle biopsy samples. We applied Biop-C to characterize three nasopharyngeal cancer solid cancer needle biopsy patient samples. We identified topologically associated domains (TADs), chromatin interaction loops, and frequently interacting regions (FIREs) at key oncogenes in nasopharyngeal cancer from the Biop-C heatmaps. We observed that the genomic features are shared at some important oncogenes, but the patients also display extensive heterogeneity at certain genomic loci. On analyzing the super enhancer landscape in nasopharyngeal cancer cell lines, we found that the super enhancers are associated with FIREs and can be linked to distal genes via chromatin loops in NPC. Taken together, our results demonstrate the utility of our Biop-C method in investigating 3D genome organization in solid cancers.

In this study we performed a multi-omics analysis comprising whole-exome sequencing (WES) and RNA sequencing (RNA-Seq) on seven breast cancer patients, consisting of three Estrogen receptor (ER) positive and four Triple negative breast cancer (TNBC) subtypes to understand the neoantigen burden in breast cancer tumor samples. We predicted both class-I and class-II human leukocyte antigen (HLA) bound neoantigens by analyzing matched tumor-normal pair of exomes. Across all the patients, we predicted 434 unique neoantigens (NeoFil) in total, affecting 237 different genes and 87% of them (n = 378) are expressed at RNA level (Neoexp). The missense mutations (87%) are the major contributor in neoantigen (Neoexp) generation, followed by frameshift (11%) and indels (2%). The neoantigens (NeoFil) were found to be positively correlated with the somatic mutations (R2 = 0.89). We also noted that the vast majority (99.98%) of the predicted neoantigens are patient specific. Overall, the current study offers significant insight into the neoantigen profile in tumor types with intermediate/low mutation burdens like breast cancer.