NEW YORK – Illumina's in-house CLIA lab has validated the use of its ExpansionHunter software to diagnose rare diseases caused by short tandem repeat (STR) expansions as part of its clinical whole-genome sequencing assay.
"For the first time, we can go after repeat expansion loci all at the same time. We don't have to try to pick and choose which expansions we're going after," Tanner Hagelstrom, laboratory director at the Illumina Clinical Services Laboratory, said in a presentation on Tuesday at the virtual American Society of Human Genetics meeting, describing its validation process.
Using 168 samples orthogonally confirmed by an outside clinical lab, Illumina researchers analyzed four classes of STR expansions associated with nine different conditions, including fragile X syndrome, myotonic dystrophy, Huntington's disease, Friedrich ataxia, spinal-bulbar muscular atrophy, and spinocerebellar ataxia. ExpansionHunter provides greater than 97 percent sensitivity and specificity, "almost approaching the standard of care," Hagelstrom said, and works across all family structures. The method is limited to loci that had a normal range within a 150-bp read length, he noted.
In about a year, the lab has used the test to find five cases where STR expansions are responsible for diseases including myotonic dystrophy, spinocerebellar ataxia, and Friedrich ataxia.
Released publicly in late 2016, the open-source ExpansionHunter software is a graph-based algorithm that helps analyze STR expansions in whole genomes generated with PCR-free library prep. Illumina most recently updated the software to a new version earlier this month. In a study published in Genome Research in 2017, the algorithm correctly identified expansions or potential expansions in the C9orf72 gene for all 212 amyotrophic lateral sclerosis patients in a study, and was 99.9 percent accurate in classifying nearly 2,800 other samples as wild-type.
It's one of several software packages available for STR expansion analysis; others include exSTRa, STRetch, and Tredparse.
Using genome graphs, ExpansionHunter portrays STR expansions as loops in between the unique flanking regions. "The reason traditional aligners have a hard time calling STR expansions is that you get a lot of variability compared to the reference [genome,]" Hagelstrom said. The variants get flagged as indels and calling them becomes too difficult.
Beyond technical difficulty, STR expansions are hard to test for from a clinical perspective. "Many labs don't do all of them, they're too rare to set up a standalone assay," Hagelstrom said.
Illumina's lab uses a minimum sequencing depth of 30x coverage, but Hagelstrom said it was probably more like 35x or 37x coverage.
Illumina's clinical strategy is to only report STR expansions that have a phenotype. They also make sure to do a visual inspection of the expanded allele and do orthogonal characterization. "We recommend having a strong incidental finding policy, which you should probably have for WGS anyway," Hagelstrom added.
The method comes with some caveats. ExpansionHunter may miss mosaic features and the validation study did not include polyalanines, since they did not have patient samples and that class of expansions requires a slightly different algorithm, he said. But polyalanines are a future direction the lab is interested in.
Illumina also offers a de novo ExpansionHunter algorithm, which is another aspect of STR expansion analysis that the lab would like to incorporate.