Allelica

The model is the engine of the core product, and the platform carries some other value. What Allelica sells is polygenic risk scores, statistical models fitted to large genotyped cohorts with ancestry specific weights, either as software a laboratory runs or as tests from Allelica's own laboratory, and none of that exists without the models. AbsoluteDx and Breast Risk Reveal combine the scores with monogenic findings and, for breast cancer, the licensed BOADICEA model.

The Allelica Precision Server also reports pharmacogenomics and carrier screening for laboratories, which is bioinformatics reporting rather than model output, so the platform is not purely the model.

Ask which results in a report come from Allelica's fitted scores, which from licensed models such as BOADICEA, and which from variant reporting.

Scores reach a laboratory or an ordering clinician, and the controls around them are not described. Laboratories run Allelica's software inside their own validated tests and sign out results, and clinicians order Allelica's tests and decide what a risk estimate means for treatment, which is review before action.

What is not published is the mechanism: which ancestry or quality conditions stop a score from being reported, how scores outside validated groups are labeled, and how a recalibrated model affects scores already issued.

Ask which quality and ancestry checks gate a reported score and how laboratories are told when a model changes.

The approach and a supplier are named without version and update discipline. Allelica describes ancestry specific polygenic risk scores developed and validated in named cohorts, software modules to discover, validate and run them, and the BOADICEA breast cancer model licensed from Cambridge Enterprise inside Breast Risk Reveal. Its scores run on Illumina and Thermo Fisher genotyping platforms through those partnerships.

No version history for the scores, recalibration schedule or notice to laboratories when a model changes is published.

Ask which score versions your laboratory would run, how updates are validated and released, and whether reports carry the version.

The chain is short and the outside model is named. Allelica builds its polygenic scores in house, licenses the BOADICEA breast cancer model from Cambridge Enterprise for Breast Risk Reveal, and offers an on premises server in which no transfer to Allelica occurs at all. Illumina and Thermo Fisher are named as platform partners.

For the cloud service, no hosting provider, subprocessor list or terms governing customer genotypes are published.

Ask for the processors behind the cloud service and the terms that govern customer data once it arrives.

Peer reviewed development and validation by the company's own scientists. In Nature Communications (2023), Allelica authors built ancestry specific coronary artery disease scores on 29,389 people from diverse cohorts and ancestry groups, reported an average odds ratio per standard deviation of 1.57 with 12 to 24 percent of people placed at high genetic risk, and, in a separate cohort of 9,691 people, found a net reclassification improvement of 13.14 percent (95 percent CI 9.23 to 17.06) for coronary artery disease and 10.70 percent for atherosclerotic cardiovascular disease among those at intermediate ten year risk. Allelica also lists a cost effectiveness study in the Journal of the American Heart Association and work in Circulation on polygenic risk and LDL cholesterol.

Allelica states that the 2026 ACC/AHA dyslipidemia guideline cites its multi ancestry study; that is a citation of the research rather than an independent evaluation of the product. No prospective study of outcomes in patients tested with Allelica's scores is published.

Ask for results from customer deployments, such as statin starts or events among people reclassified by the score.

Nothing published describes how patient genetic data moves through the system. The only privacy document is a 2020 website policy that does not address genotype data, retention, research use or whether customer data trains or recalibrates the scores. The on premises Precision Server can keep data inside the laboratory, but no terms are published for the cloud service.

Safety engineering for the scores, meaning how a faulty score would be detected and handled, is not published.

Ask how long genotypes and results are kept in the cloud service, whether customer data refines Allelica's models, and what de identified data Allelica retains.

No status is stated and no privacy document reaches the product. The privacy policy Allelica publishes, last updated 2 January 2020, covers its website and general service, does not mention HIPAA, genetic data or clinical testing, and predates the clinical tests it now sells from its own laboratory. No business associate position is stated for the software laboratories run on patient genotypes, and no notice of privacy practices is published for patients tested in Allelica's laboratory.

Ask for Allelica's HIPAA role for each product, the notice of privacy practices that covers tested patients, and the business associate agreement offered with the hosted software.

Nothing independent covers information security. Allelica holds ISO 13485 and ISO 9001, which certify quality management for medical device and general processes, and a CLIA certificate for its laboratory; none is an audit of how it runs security. No SOC 2 report, HITRUST certification, ISO/IEC 27001 certificate, penetration testing statement or trust center is published, and the website privacy policy says only that commercially acceptable means protect data.

Ask for a security attestation covering the cloud service, with scope and date.

No device claim is made and the tests are positioned as laboratory developed tests. Allelica says it launched its ancestry specific polygenic risk score tests as laboratory developed tests from its CLIA certified laboratory, and that the Precision Server is designed for laboratories working under CLIA. It holds ISO 13485 certification for medical device quality management and ISO 9001, which describe how it builds products rather than a clearance.

No FDA clearance or CE marking is claimed for any score, and the regulatory position of the software when another laboratory reports its scores is not stated.

Ask for Allelica's regulatory position for the software sold to laboratories and whether any score is planned for FDA submission.

Ancestry is built into the method and the group results are not set out on Allelica's pages. Polygenic scores usually perform worse outside European ancestry, and Allelica's response is ancestry specific scoring: its Nature Communications paper (2023) develops separate scores across genetic ancestry groups in 29,389 people and reports performance as an average odds ratio per standard deviation of 1.57 across them. Allelica describes the paper as showing clinical utility across ancestry groups.

The per group figures, which show where a score is weakest, are not presented on the site or in the paper's summary, and no change control or monitoring process for the models is published.

Ask for odds ratios and calibration by ancestry group, and for how Allelica reports a score for a person whose ancestry falls between reference groups.

Nothing published says what happens when a score is wrong. Allelica publishes discrimination and reclassification figures for its coronary artery disease scores, which describe how well the scores separate risk rather than how often a person is misplaced at a decision threshold. No error rate at reporting thresholds, warranty, remediation commitment or correction route for a tested patient is published, and no notice of privacy practices sets out a patient's rights over Allelica's laboratory results.

Ask how many people the score moves across the treatment threshold in error, and what Allelica commits to when a reported score is later corrected.

No integration evidence is published. Allelica's software runs inside laboratories' bioinformatics workflows and reports through them, and its tests return reports to ordering clinicians, but no record system integration, standard, marketplace listing or live site is described for either route. Named institutions working with Allelica are research and implementation partners rather than integration references.

Ask how scores reach the chart at your institution, whether as discrete results or as a document, and for a site where that runs today.

Options are stated, with the cloud location left open. Allelica offers its software as a cloud service and as the on premises Allelica Precision Server, which a laboratory runs in its own environment so genotypes need not leave it. That gives a buyer a residency choice by deployment.

For the cloud service, no hosting provider, region or tenant isolation is described, and Allelica operates in the United States and Europe without saying where each customer's data rests.

Ask where the cloud service is hosted for US customers, whether European data stays in Europe, and how one laboratory's data is separated from another's.

No price is published and how the products are sold can be established. Laboratories and health systems license the software, as a cloud service or as the on premises Allelica Precision Server, and clinicians order Allelica's tests from its laboratory. The site lists the modules and tests without a license fee, per sample rate or test price, and no insurance position for the tests is stated.

Ask whether the software is priced per sample, per seat or by annual license, and the price and billing route for each clinical test.

Coverage is named with validation behind part of it. The coronary artery disease scores are validated across ancestry groups and for reclassifying people at intermediate cardiovascular risk, which is the preventive cardiology setting the guideline discussion concerns. Breast cancer risk is offered through Breast Risk Reveal and AbsoluteDx, and Allelica lists scores for other conditions, including Alzheimer's disease, without published validation for each on its pages.

Ask which conditions and ancestry groups each score is validated for, and how results are reported for people outside them.