05 Oct Getting a Genetic Diagnosis at the UDN
By Joy Cogan | October 5, 2022
Some people call my role a “genomicist”. I’m not the type of doctor you would ever meet at a clinic appointment (not an MD); I am a PhD researcher. My degree is in Biochemistry. I search through the very detailed code that makes you, well You. Why is it that the UDN needs someone like me at the Clinical Site to do this instead of simply outsourcing it to the genetics laboratory where you got your WES/WGS CLIA test report from?
The UDN system has a two-tiered approach to tracking down a genetic diagnosis. First, the typical genetic test that would also be available clinically (with a few tweaks to make it more sensitive to rare diseases), is called WES or WGS. We run our WGS at read depth which is far greater than what you’d normally get. This ensures that we have the highest quality test data to work off of when we do our deep dive. CLIA labs, such as our lab at Baylor Genetics, are compelled by their licensure to only put things on the report that are known in the literature & the NIH database ClinVar to be pathogenic, likely pathogenic, or a VUS. This is why most of the time, unsurprisingly, the Initial CLIA report is blank. (We expect that in 90% of the cases the report will not reveal the answer to us. The majority of patients accepted into the network have already had extensive genetic testing and seen a dozen specialists – if it was easy to find, it would have been found already!) Since they are part of our Network of researchers, the CLIA lab also provides us with another report, called Table 5, a research-level report that is for the research personnel’s eyes only with their guesses of other variants they saw that might be of interest but do not meet the criteria to put on the Initial report. Second, the UDN Clinical Sites each have a pipeline to dig deeper – do a deep dive so-to-speak. I’ll go into greater detail about this…
Your phenotype (symptom-set) has been coded into our database by the clinicians in a code language called Human Phenotype Ontology. This is a standard dictionary of medical terminology that uses a decision tree of descriptors for the purpose of forcing all clinicians to use the same word for each symptom, without variation. Example: microcephaly (HP:0000252) instead of small head. Now those coded words are given a number and computers can readily mine the data across patients to look for trends. When the Clinical Site does the deep dive, we input those codes into the software that reads your DNA (example: at Vanderbilt, I use Emedgene and Omicia Opal) so that it can output the variants of interest that align with the symptoms we entered. Even if these symptoms are in mice, instead of humans (via a software developed by UDN scientists called MARRVEL)! The remarkable thing about this deep dive is that it lets us see the rare variants through the lens of these “filters”. Our researchers evaluate how rare a variant is across multiple factors (filters): how common is it vs. the general population (UK biobank; GnomAD); if it’s a missense variant, how bad are the domino effects (CADD; Missense-Z); is this a “loss of function” variant (LoF pLI) – which tells us if the unknown condition is likely to take on a dominant or recessive inheritance; is this a part of the DNA code that over time has proven in both humans and animals that it is not tolerant to change (GERP). The mixture of these scores lets us prioritize which novel gene variants might be the cause of this potentially first-time-in-human disease (novel disease).
I say “our researchers” because it is not just me, it’s often a room of 20 people – a joint conference of medical biologists and informaticians – who have put 40 hours into combing through you or your child’s DNA for these answers. We also look at through some special lenses at our site that allow us to compare your DNA to other sets of data: a sample of the overall patients at Vanderbilt lets us see if there is a matching case here in our own city, a cadaver database lets us know how the gene behaves in your liver vs. your brain, and our structural biologists can tell us what the potential impact on the protein that variant might cause in your body’s ability to do its job (this looks like a picture of one of those old telephone cords… all curly, in the computer model!).
Once we’ve identified variants we want to explore further, we double check these at the CLIA lab by a different method – Sanger sequencing. The Sanger sequencing report is called the Final Report. The Final report reflects the approximately 1-8 hypotheses we have that we want to look into further with match-making or functional testing. These are “variants of uncertain significance (VUS)” and are not the answer but rather a list of our favorite possibilities.
The UDN funding allows us to do many research tests (functional tests) on your samples to try to find the answer. Things that no clinic is able to offer, as most of the time it cannot be billed to insurance. Examples: segregation studies, cDNA studies, realtime PCR, RNAseq, long read sequencing, etc. We can even look for chromosomal rearrangements and structural variants. Lots of uber cool, nerdy science.
The UDN is truly nothing less than one of the best scientific collaborations I’ve had the opportunity to contribute to. “Collaboration” is the heart of the UDN.
The UDN thus far has identified over 50 novel diseases, which now contribute to the now over 9,500 known rare diseases. So collectively, “rare” is far less rare than you think. If 10% of the American population has a rare disease then it is fairly as likely that you know someone with a rare disease as it is that you know someone with Type 2 diabetes.
You can learn more about genetic sequencing in the UDN PEER’s Sequencing Guide for families on the UDN Resources page. A guide about RNA Sequencing is also available on the same page: