What Actually Happens When You Try to Put Genomics Into Routine Care
Clinical genomics isn't a neat package you can just adopt and move on from. It's a series of workflows, interpretations, and conversations that most hospitals have never had to coordinate before. I spent years watching institutions try to graft genetic testing onto standard clinical practice, and the failures usually came down to one thing: people assumed the data would speak for itself. It doesn't. The variant gets returned, then nobody in the clinic knows what to do with it, and the result sits in a report that never gets acted on. The core idea behind Valerie Sandler's work is straightforward on paper. She focused on building the infrastructure that makes it possible to actually use genomic data in a clinical setting rather than letting it collect dust in a research database. That means standardized interpretation pipelines, clear reporting formats, and pathways that connect a lab result to an attending physician who knows what action to take. The transformation in patient care comes from closing that loop. I remember running into a real problem when we first tried to implement whole-exome sequencing for undiagnosed pediatric cases. The lab would return results in about three weeks, but the variant classification reports used different criteria than what our clinical team was familiar with. A VUS one lab called likely pathogenic, another lab would downgrade it. We were seeing inconsistent follow-up because the physicians couldn't trust the severity assignments. The workaround was relatively simple but took months to set up: we created a mandatory pre-test counseling session that outlined exactly what the test could and couldn't tell us, and we required a clinical genetics review before any result went back to the ordering provider. After that, the variance in how results were handled dropped significantly.
The technical side of clinical genomics hinges on a few specific components. You need sequencing infrastructure, which is typically either targeted gene panels or exome or genome sequencing. Then you need a bioinformatics pipeline that calls variants against a reference genome, annotates them with population frequency data from sources like gnomAD, checks ClinVar and other databases for known clinical significance, and filters out benign polymorphisms. The hardest part isn't the sequencing itself. It's the interpretation. That's where most programs stall out. Variant interpretation follows the ACMG/AMP guidelines, which break classifications into five tiers: pathogenic, likely pathogenic, variant of uncertain significance, likely benign, and benign. The problem is that the guidelines leave significant room for human judgment. A single missense variant can end up in different categories depending on which bioinformatics tools the analyst weights more heavily. I've seen the same BRCA1 variant classified differently at two different reference labs simply because they used different splice prediction algorithms. That's not a hypothetical edge case. It happened to me directly with a family we were counseling, and it delayed treatment decisions by six weeks while we sought a third opinion. Here's what most people don't understand about clinical genomics: the 20 to 30 percent of cases that return a variant of uncertain significance are the ones that actually determine whether a program succeeds or fails. A definitive pathogenic result is clean. It validates the test. But a VUS creates a cascade of follow-up questions, repeat testing, family studies, and sometimes misguided clinical decisions based on incomplete information. Programs that don't build strong genetic counseling capacity around VUS management tend to accumulate a backlog of unresolved cases that erode clinician confidence over time.
The practical workflow for implementing clinical genomics in a care setting looks like this. First, you define the clinical indications. Not every patient needs a genome. You start with specific phenotypes: rare diseases in pediatric populations, oncology cases where tumor sequencing changes treatment selection, pharmacogenomics for patients on complex medication regimens. Then you build the lab partnership or in-house capability. This is where most timelines get because selecting a CLIA-certified laboratory and negotiating turn-around times and reporting standards takes longer than expected. A typical contract phase runs four to eight months. Once the testing pipeline is operational, you need EHR integration. The genomic data has to land somewhere clinicians can actually find it. I've seen too many programs where the sequencing results went into a separate portal that attending physicians never logged into. The data existed but was functionally invisible. The fix is embedding variant reports directly into the patient's chart with structured data fields that trigger clinical decision support alerts when a actionable variant is identified. This usually requires working with your EHR vendor, and the integration can add another three to six months depending on your platform. Pharmacogenomics is probably the highest-return application for most clinical settings. Drug-gene interactions are well established for CYP2D6, CYP2C19, TPMT, and several others. Testing these before prescribing can prevent adverse drug reactions and improve therapeutic outcomes. The turn-around time is shorter, the interpretation is more standardized, and clinicians accept the results more readily because the evidence base is solid. I'd recommend starting any clinical genomics program with pharmacogenomics as a pilot. It builds institutional comfort with genomic data while delivering measurable outcomes quickly.
Get the Full Details
There are real limitations worth being honest about. Clinical genomics currently identifies a causative variant in roughly 25 to 40 percent of appropriately selected rare disease cases. That means 60 to 75 percent of patients who undergo testing get no definitive answer. This isn't a failure of the technology. It's a limitation of our current knowledge. We simply don't have enough functional data to interpret many variants, and some diseases are caused by mechanisms we still can't detect with standard sequencing, like structural variants below the resolution threshold or epigenetic disorders. Programs that don't manage patient expectations around these limitations will face distress and disillusionment. Data privacy and genetic discrimination concerns are real even with GINA protections in place. GINA covers health insurance and employment but doesn't protect against life insurance, disability insurance, or long-term care insurance. Patients need to know this before testing. I've had to counsel families who declined genomic testing solely because a relative had previously been denied life insurance after incidental genetic findings surfaced. That's a legal gap that no amount of clinical infrastructure can fix. The cost landscape has improved dramatically. Whole-exome sequencing now runs approximately $400 to $800 per sample in most clinical laboratories, and whole-genome sequencing is approaching $600 to $1,000. Reimbursement is still uneven. Some payers cover exome sequencing for specific pediatric indications, while others require extensive prior authorization that delays results by weeks. Building a reimbursement strategy should be part of your implementation plan from month one, not an afterthought.
If you're considering launching or improving a clinical genomics program, the most practical first step is assembling a multidisciplinary team that includes a medical geneticist, a genetics counselor, a bioinformatician, and an EHR integration specialist. You don't need all four on day one, but you can't skip any of them long term. The next step is picking a narrow clinical indication and running a pilot with 50 to 100 cases before scaling. This lets you identify workflow bottlenecks, refine your reporting templates, and train your clinicians on how to act on genomic results without committing institutional resources to a system that hasn't been stress-tested. The transformation in patient care that clinical genomics enables is real but incremental. It doesn't replace thorough clinical evaluation or replace the diagnostician's intuition. What it does is add a layer of information that was previously unavailable, and that information changes management decisions in a meaningful minority of cases. The value isn't in the sequencing. It's in what happens after the sequence data comes back.
