Deconvoluting complex genomic structural variations in metastatic tumors
Guest Author: Stephanie Greer, Stanford Medicine, Ji Research Group
Genomic rearrangements in oncogenesis
Cancer rearrangements that involve large structural changes in the human genome are frequent somatic events seen across tumor types and play an important role for driving oncogenesis. Determining the precise structure of a cancer rearrangement has implications for tumor evolution and oncogenic gain-of-function events as therapeutic targets. Oftentimes, cancer rearrangements consist of multiple coincident structural variations (deletions, duplications, inversions, translocations etc.) in the same gene or genomic region. As a result, cancer structural variants (SVs) are extremely complex and difficult to resolve. Although these types of genome changes are known disease driving candidates, there are no reliable, cost-effective methods available to resolve such complex chromosomal rearrangements at the sequence level.
Unique SV breakpoints identified in metastatic tumor samples
In our recent study published in Genome Medicine (Greer _et al. _2017), we used Linked-Read sequencing to conduct a comparative analysis of two ovarian metastases originating from a primary, diffuse gastric tumor. We detected what appeared to be tight clusters of SV breakpoints in the genomic region harboring the candidate oncogene, FGFR2. Upon closer inspection, it became apparent that the SV breakpoints were unique between the two surgically excised metastatic samples.
Linked-Reads resolved complex SVs in megabase-scale haplotypes
Our study determined that traditional WGS data was inadequate for our purposes. With Linked-Reads, however, we were able to deconvolute the associations among rearranged regions of the genome from these tumor samples. We developed a tool to phase and resolve complex SVs. Our method leveraged the identification of high molecular weight DNA molecules (>20 kb) that span the SV breakpoint(s). These "SV-specific" barcodes were used to phase SVs with respect to one another by placing them into the context of existing megabase-scale haplotypes. This approach also enabled a streamlined de novo assembly approach—using only those barcoded sequence reads that belong to the SV events, we assembled more accurate and complete genome structure at the SV sites.
With barcoded Linked-Reads and a series of novel bioinformatics tools we conceived, SV events were phased within each ovarian metastasis and determined to be in cis with one another, thus occurring on the same haplotype. Furthermore, we found that although all of the SV events were completely distinct between the metastases, the same haplotype was actually affected in both metastases. A haplotype-specific de novo assembly of the SV genomic regions generated large (kb) contigs that supported our putative rearranged structure. Overall, leveraging the power of Linked-Reads enabled us to resolve the complete structure of cancer genomic rearrangements.
Additional Resources
- Learn more about the Chromium™ Genome Solution.
- Read Chromium™ Genome Solution application notes
- Check out our data analysis tools
- Download Chromium™ Genome Solution datasets
Read more about using Linked-Reads for strucural variant detection on the blog