Joshua Law
PhD student
Supervisors: Associate Professor Tu Nguyen Dumont, Dr Jason Steen, Professor Paul Lacaze.
What I’m Working On: Changing the story DNA tells about cancer and heart disease risk
Roughly one in 100 people carry a hidden genetic variant that increases their risk for developing cancer or heart disease. They may be completely healthy, but they are unaware of their elevated risk for developing disease at a younger age. What if we could change that?
My project is part of a national research effort called DNA screen. For the first time in Australia, we offered free genetic testing to 10,000 healthy Australians. The goal of this study is to identify people carrying genetic variants that increase their risk for disease and offer them the chance to take preventative measures using saliva DNA.
We've already identified many at-risk individuals by spotting tiny genetic differences- for example, a single letter in their DNA that's swapped. However, we're currently missing a very important class of genetic variants known as copy number variants. These are large chunks of DNA are either missing or duplicated.
Based on our current estimates, we expect a dozen people in our study may carry a copy number variant that increases their risk for disease. However, we're currently not able to find these individuals due to a limitation in our genetic tests.
Imagine your DNA as a massive book containing billions of DNA letters in it. Our test works by reading each page in this book multiple times to find copy-number variants. We observe how many times each page is being read. If a page is read more often than expected, it might indicate a duplicated section. But here lies the limitation of our tests: they don't read each page evenly. This means some pages are read more often than others, simply by chance, and this random variation occurs because our test only looks at a small number of pages relevant for disease risk. So it's hard for us to tell if a tiny fluctuation in our reading is due to a true biological signal indicating a copy number variant, or simply just technical noise.
To tackle this problem, I developed a computational framework that best detects copy-number variants in my data set. First, I ran an algorithm that scanned the raw data and detected over 1000 candidate copy number variants. Then I simulated real copy number variants and used them to filter candidates that don't match expected patterns.
Then I selected the most promising candidates and sent them for independent testing. This then formed a reference set, which I can then use to fine tune other detection algorithms and help me put together the best combination of algorithms that detects copy-number variants most accurately in my data set.
Ultimately, I hope that I'm able to find more individuals carrying hidden genetic variants, and most importantly, offer them the chance to take action before disease takes hold. In the end, it's not about how you read the DNA - it's about changing the story it tells.
*This story is based on Joshua’s 2025 Three-Minute Thesis entry.