With the advent of next generation sequencers (NGS) that provide sequencing at a substantially lower cost, the development SNPs at the level of whole genomes can be done in a single laboratory. However, genome structural variation including large insertions and deletions, and chromosomal reciprocal translocations has not yet been focused due to the limitation of re-sequencing methods as genome structures rely to that of a known reference genome. For an improved detection of the structural variations after deep re-sequencing of the Glycine soja accession CS-14, we de novo assembled the whole paired-end reads (W-contigs). After the de novo assembly, the paired-end reads that did not match the reference genome of Williams 82 were retrieved and de novo assembled them (U-contigs). We then predicted structural variation candidates. For predicting the function of the structural variation candidates, we compared those structural variation candidates with SwissProt DB using BLASTX. Most of them were matched with transposable element related proteins or stress tolerance related proteins (Table 1). We designed 24 primers for all candidates and tested in CS-14 and Williams 82 for validation. As a result, the DNA polymorphism was observed between CS-14 and Williams 82 in the three primer sets, CS14IC10, CS14IC12 and CS14IC15, with the expected size of the PCR product . For further validation, we sequenced the DNA band amplified by CS14IC15, and its sequences were aligned well against the Williams 82 and CS-14 contig. Especially, IC15R-CS14 was aligned in the predicted insertion region, consequently, this sequenced region would indicate structural variation. The other primer sets did not amplified either because they were designed for an amplifying long genomic region or because of the fragmented template DNA