Structural variants, particularly distant genomic translocations, are difficult to identify despite their fundamental importance in disease. Current short-read genomic approaches suffer from high rates of false positives because of the massive multiple-testing problem, and cannot detect variants in repetitive genomic regions. The 10X Genomics platform generates barcoded short-reads from long genomic DNA fragments, which can then be clustered in silico to generate read clouds identifying the original large DNA fragments. I will discuss new methods we are developing to identify structural variation from 10X data based on similar barcode patterns observed between distant genomic locations. These new methods dramatically improve sensitivity and specificity over standard short-read methods, and allow us to better understand complex structural variation and haplotype context. I will also briefly discuss application of these methods to gain insight into the evolution of structural variants within cancer.