Do You Need a Doctor's Order for Genetic Testing?

In the United States, most clinical genetic tests require an order from a licensed provider. That requirement is a paperwork gate. It is not a gatekeeping judgment about whether you deserve your own data. Labs operating under CLIA release results to the ordering provider, and many consumer-facing sequencing services satisfy that by routing your request through an independent physician network that reviews and signs the requisition. A small number of tests are authorized for direct consumer purchase with no provider involved. And if you searched this phrase because a court is involved, that is an entirely different thing: a judge ordering parentage testing under state family law, which has nothing to do with the clinical requisition described below.

The two things called “an order for genetic testing”

A court order for genetic testing is a judicial instruction compelling a party to provide a buccal swab or blood sample for parentage determination, with chain-of-custody handling by an AABB-accredited lab. Home kits are generally not admissible for that purpose. If that is your situation, you want a family law attorney. This page will not help.

A clinical order is a test requisition: a form or API call identifying the patient, the specimen, the test code, an indication, and a provider NPI. It authorizes the lab to accept your sample and release results. That is the document people mean when they ask how to order sequencing.

When a provider order is legally required, and when it is not

Three layers stack here.

CLIA governs the lab. A CLIA-certified lab reporting results intended for diagnosis or health assessment reports to an authorized person, which in most states means a licensed provider. Research sequencing performed in a non-CLIA lab is outside this frame entirely, which is why research-grade whole-genome data can be delivered to you without a physician anywhere in the loop, and also why no clinician should act on it without confirmation.

State law sets the floor. Most states now permit direct-access testing for many lab tests. New York has been the strictest about requiring an ordering provider and about which labs may accept New York specimens at all. State rules change, so check the current statute for your state before assuming a workflow will run.

FDA regulates specific consumer devices and reports. A handful of consumer genetic health reports are authorized for direct purchase, notably a narrow panel of founder variants rather than a full gene analysis. The authorized report covers three BRCA1/BRCA2 Ashkenazi founder variants out of thousands of known pathogenic variants in those two genes, which is the kind of detail that matters if you are trying to reason about what a negative result means.

In practice, the common resolution is a physician-order network. You complete an intake, a licensed physician in your state reviews it and signs the requisition, and the lab runs the test. The physician is real and is legally the ordering provider. For most healthy-person baseline sequencing, this is the path we would use. It is fast, it keeps the result inside a CLIA-regulated chain, and it does not require you to persuade a primary care physician that a genome is a reasonable thing to want.

How to place the order and what to specify

Ordering is easy. Ordering something worth analyzing takes four decisions.

Assay. Whole-genome sequencing, not an exome, and not a genotyping array. An array interrogates roughly a million fixed positions and cannot see anything not on the chip. An exome covers about 1 to 2 percent of the genome with uneven capture at GC-rich first exons. WGS gives you uniform coverage, better copy-number resolution, and the intronic and regulatory sequence you will want in five years when the interpretation catches up. The cost argument that once favored targeted assays has mostly evaporated as per-genome sequencing cost fell by orders of magnitude from the mid-2000s trajectory ¹.

Depth and library. Ask for 30x mean coverage minimum on a PCR-free library, 2x150 bp reads. PCR-free matters for even coverage in homopolymers and GC extremes and for cleaner duplicate rates. Ask what fraction of the genome is covered at 10x or better. Below about 90 percent, ask why.

Reference build. Confirm GRCh38 with alt contigs handled explicitly, or ask for unaligned data and do it yourself. If the lab hands you hg19 output, you will spend real time on liftover artifacts around segmental duplications.

Deliverable. This is the decision people skip and regret.

The files to demand

Specify in writing, before the sample ships, that you receive:

• FASTQ or unaligned BAM with original base qualities. Without these you cannot realign to a new reference, and you are stuck with whatever pipeline the lab ran in 2026.
• CRAM or BAM aligned to a named reference, with the reference URI recorded in the header.
• gVCF, alongside the filtered VCF. The filtered VCF alone is not enough. A gVCF distinguishes homozygous reference from no-coverage, which is the difference between “you do not carry this variant” and “we could not see that position.”
• A coverage report and the pipeline version string.

A 30x human genome is roughly 90 to 120 GB as FASTQ, about 40 to 60 GB as a CRAM against GRCh38. Plan storage and checksum everything on arrival with md5sum -c.

A reasonable local pass, if you want to rebuild from raw reads:

bwa-mem2 mem -t 32 -R '@RG\tID:L1\tSM:SAMPLE\tPL:ILLUMINA' \
  GRCh38.fa r1.fq.gz r2.fq.gz | samtools sort -@8 -o s.bam
samtools index s.bam
run_deepvariant --model_type=WGS --ref=GRCh38.fa \
  --reads=s.bam --output_vcf=s.vcf.gz --output_gvcf=s.g.vcf.gz --num_shards=32
bcftools view -f PASS s.vcf.gz | \
  vep --cache --assembly GRCh38 --everything --vcf -o annotated.vcf

Then intersect against ClinVar with bcftools annotate -a clinvar.vcf.gz, and use gnomAD allele frequencies to discard the long tail of rare-but-benign. Structural variants need a separate caller (Manta or GATK-SV), and repeat expansions need ExpansionHunter with a supplied variant catalog. Short-read WGS will still miss some of these, and a clean VCF is not a clean genome.

What an order does not give you

It does not give you an interpretation you can act on. Diagnostic yield for WGS in people with an actual clinical presentation runs around 40 percent across clinical categories in large rare-disease cohorts ², and that is in symptomatic patients with phenotype-driven filtering. In an asymptomatic person, the prior is far lower and the dominant output is variants of uncertain significance.

It also does not settle what you want told to you. Secondary and incidental findings, carrier status with implications for relatives, and the consent problem for family members who share your variants are all decisions to make before the report exists rather than after ³.

Any variant that looks clinically meaningful needs orthogonal confirmation in a CLIA/CAP lab and interpretation by a medical geneticist or certified genetic counselor before it informs any medical decision. Research-grade data is not a diagnosis, and neither is an annotated VCF.

Questions people also ask

Do you need a doctor’s order for genetic testing? For most clinical genetic tests in the US, yes, but the order can come from an independent physician network rather than your own physician. A narrow set of consumer reports are FDA-authorized for direct purchase, and research-grade sequencing performed outside a clinical reporting framework does not require an order.

How do you order genetic testing? Choose the assay and depth, then pick a lab and confirm the deliverable file formats in writing. Complete the intake so a licensed provider can sign the requisition, then submit saliva or a blood draw. Expect three to eight weeks from sample receipt to data release for WGS.

What kind of doctor orders genetic testing? Any licensed physician, nurse practitioner, or physician assistant can order it within their scope. Medical geneticists and genetic counselors are the specialists who interpret results, and a genetic counselor is the person to see first if you have a family history driving the question.

Can I analyze the data myself? Yes, and you should, if you have the skills. Run your own alignment and calling, keep the raw reads, and treat every finding as a hypothesis until a clinical lab confirms it.

Footnotes

1. Simon T. Bennett, C.L. Barnes, Anthony J. Cox, et al. Toward the $1000 Human Genome. Pharmacogenomics, 2005. https://doi.org/10.1517/14622416.6.4.373 ↩

2. Henrik Stranneheim, Kristina Lagerstedt‐Robinson, Måns Magnusson, et al. Integration of whole genome sequencing into a healthcare setting: high diagnostic rates across multiple clinical entities in 3219 rare disease patients. Genome Medicine, 2021. https://doi.org/10.1186/s13073-021-00855-5 ↩

3. Wim Pinxten, Heidi Howard. Ethical issues raised by whole genome sequencing. Best Practice & Research Clinical Gastroenterology, 2014. https://doi.org/10.1016/j.bpg.2014.02.004 ↩