Abstract Introduction Genomic analysis of multiple myeloma has advanced greatly in recent years, but traditional short-read sequencing has limited the power to examine complex or linked events across the genome. Linking markers together into large phase-blocks, and examining the interaction of different sequencing modalities, allows complex genomic and epigenomic states to be integrated to generate true multiomic haplotypes at the chromosomal level. Methods Fourteen multiple myeloma patient-derived xenografts were used to generate multiomic data including short- (Illumina) and long-read (PacBio) whole genome sequencing (WGS, average depths 92x and 16x) to identify single nucleotide variations (SNVs), copy number (CN) abnormalities, structural variation (SV), and DNA methylation, as well as expression (124 million reads), chromatin states (Cut10M reads, H3K27ac, H3K4me1, H3K4me3, H3K9me3, H3K27me3, and H3K36me3), and Micro-C/LinkPrep (Dovetail Genomics; 900M reads) and HiChIP (Dovetail, H3K27ac mark) to identify 3D chromatin architecture/folding and enhancer-target gene interactions. Micro-C/LinkPrep interaction reads were also used for chromosome-scale haplotype phasing, onto which SNVs, SVs, CNVs, DNA methylation, expression, and chromatin marks were super-imposed based on heterozygous SNPs. Data integration resulted in haplotype-phased somatic events for genomic architecture, SNVs, SVs, epigenomic states, and expression markers. Results For the first time, complete haplotype-resolved assemblies at the chromosomal level have been generated in multiple myeloma. A 100x coverage genome Micro-C/LinkPrep libraries we were able to phase the long and short arms of each chromosome, resulting in an average 96.3% haplotype phasing per autosome – thereby generating chromosome length haplotypes up to 241.9 Mb in length. In comparison, HiFi long-read (15 kb reads) WGS (PacBio) was able to generate phase-blocks of up to 4.6 Mb (median 44.1 kb). We examined the prevalence of somatic mutations on each haplotype across all chromosomes to determine if one parental haplotype was more likely to be mutated than the other. In one sample, of 19,306 mutations 56% were phased to their respective haplotypes and were generally equally distributed across both haplotypes. Haplotype-specific interaction heatmaps were generated allowing us to examine the interaction of complex SVs across chromosomes. In one sample, we identified a primary t(11;14) and additional SV events linked to the t(11;14) including a t(3;14), t(11;17), and t(3;17) which created a cyclical pattern. Using the chromosome-scale haplotype maps we were able to determine the proportion of interacting reads for the four possible haplotype combinations between any two chromosomes. We found that 70-83.5% of reads support specific combinations of haplotype interactions, and that all four SV events were linked and involved the same haplotypes on each chromosome. The pattern of interactions combined with breakpoint analysis in this case indicated a four-way complex reciprocal translocation between chromosomes 3,11,14 and 17. Integration of epigenetic data in this complex SV, including DNA methylation (HiFi reads) and super-enhancer histone marks (CUT14) breakpoint as well as increased H3K27ac marks on the same haplotype, indicating spreading of the activating broad domain from the IGH super-enhancer on chromosome 14. Equally, the hypomethylated DNA marks at the IGH promoter are spread to chromosome 3, via the t(3;14), resulting in over-expression of the proto-oncogene SKIL. The same is true for the t(3;17), which shows hypomethylation on both sides of the breakpoint, compared to the non-translocated allele. Overall, this exemplifies the underlying intricacy and impact of complex SVs across multiple linked chromosomes and their epigenetic states. Conclusion We have generated the first chromosome scale haplotype-resolved genomes in multiple myeloma and integrated them with epigenetic states. We have shown that it is possible to identify interactions across chromosomes to resolve complex SVs as well as their epigenomic consequences to understand the intricate nature of how the genome is organized.
Becker et al. (Mon,) studied this question.