Telling Lab Microservices Not to Repeat a Test Step Order Tracking for Genomic Testing

This application covers the software that runs a genomic test order from request to final report. An order is broken into tasks that many separate software services carry out, the system strips out tasks that two test templates would both require, and when one service starts a task it tells the others to stand down so nothing is done twice. Tempus AI, Inc. filed it, naming Charles Jaros, Isaiah D. Simpson, Elle Moore, Thomas Steinmetz and Robert Tell as inventors, and it is a divisional of Australian application 2020313915. The title, “Adaptive order fulfillment and tracking methods and systems”, is a fair description of what the claims cover.

The Problem

A cancer genomic test is not one step but dozens. A physician’s order leads to receipt of a tissue or blood sample, slide preparation, extraction of DNA or RNA, sequencing, computer analysis to find genetic variants, interpretation, and finally a report. Different tests share many of those steps, and different teams and software tools own different pieces.

The specification describes the target as complex medical testing order processing: adaptive systems that build a customised order made of many items, have many system resources complete them, produce reports, and show order status in real time and historically. It gives examples of items such as an accession task that records receipt of a physical specimen, a variant call task that records completion of a bioinformatics pipeline, and a variant characterisation task.

Where several templates apply to one patient, for example a DNA test and an RNA test, the steps overlap. Without coordination the same step can be queued twice, and with many independent services able to run the same kind of task, two of them may pick up the same job. The claims go at exactly those two sources of wasted work.

What This Invention Does

Claim 1 is a genomic test processing system with an order management engine connected electronically to several order processing engines. From an order request, the order management engine generates an order made of tasks to be performed by a number of microservices. At least some tasks come from a first template and a second template held in a template database. When it builds the order, it identifies duplicate tasks across the two templates and eliminates execution of the duplicates, to reduce redundant processing.

The order processing engines come in three roles. A receiving engine takes in notifications that tasks have been completed. An execution engine runs application code to manage the tasks, telling microservices when tasks should be performed. When a microservice reports an “in-progress” notification that it has started a task, the execution engine broadcasts a suppression message to the other microservices that could do the same task, so duplicate execution is prevented. A broadcasting engine sends out task-completion notifications. Throughout, the order management engine tracks each task’s status and decides the order is complete only when every task has met its completion status. The claim ends by requiring that one of the processing engines generates a next-generation sequencing report from those tracked statuses.

Claim 16 is the same thing as a method. Claims 2 to 15 then tie the generic engine to real laboratory workflows. For DNA, the steps listed include scraping a prepared FFPE (formalin-fixed paraffin-embedded) slide to collect tissue, isolating DNA, amplifying, sequencing, mapping to a human reference genome, identifying variants and generating a report. For RNA, the claims replace variant calling with measuring the abundance of mapped sequences, and one claim adds identifying fusions. Further dependent claims cover tumor-normal matched sequencing in which one engine handles the normal specimen and another handles the tumor specimen (claims 9 and 10), liquid biopsy using cell-free DNA (claim 11), a tumor mutational burden score together with a microsatellite instability score (claim 12), blood specimens with an instability score (claim 13), and copy number alterations (claim 14). Claim 10 puts the wet laboratory steps at a first geographic location and the mapping, variant identification and reporting in a cloud-based architecture.

Claims 31 and 32 are alternative independent claims with a different wording of the same idea. They speak of order templates that specify items for specific order types, the system selecting templates and combining them into an order map, removing duplicate items, an execution engine that advances the order through a sequence of steps to a final state, and a retransmitted “in progress” notification that makes other microservices “stand down” to avoid item duplication.

The specification is mainly about architecture and gives no throughput, turnaround time or cost figures for the claimed arrangement.

Key Features

• Template merging with duplicate removal. When one order draws on two test templates, the order management engine finds the tasks they share and runs each only once.
• Suppression messages. When one microservice starts a task, the execution engine tells every other service able to do it to stand down, so the same work is not done in parallel.
• Three engine roles. A receiving engine takes completion notices, an execution engine drives the tasks, and a broadcasting engine announces completion, which separates listening, acting and announcing.
• Order-level completion tracking. The order is finished only when every task has satisfied its completion status, and the final output is a next-generation sequencing report generated from those statuses.
• Parallel DNA and RNA handling. In claim 19 and its dependents, separate engines process DNA and RNA from the same specimen at the same time.
• Split between lab and cloud. The specification describes sequence data, such as base call files, being held in Amazon Web Services cloud storage, and claim 10 places the computing steps in a cloud-based architecture.

Who Is Behind It

Tempus AI, Inc. is a Chicago-based health technology company that applies data and artificial intelligence to precision medicine, mainly in cancer. Its Wikipedia entry gives its founder as Eric Lefkofsky, its founding year as 2015, and its Nasdaq listing as 14 June 2024 under the ticker TEM, and its own site, tempus.com, describes the company as having built a large library of clinical and molecular data and an operating system to make that data useful, starting with cancer. The Australian agent is Davies Collison Cave of Melbourne.

On priority, the specification is explicit. Paragraph 1 states that the complete specification of Australian application 2020313915 is incorporated by reference, and that this application claims the benefit of US provisional patent application 62/873,693, filed 12 July 2019 and titled “Adaptive Order Fulfillment and Tracking Methods and Systems”. The same paragraph also lists a related family, US application 16/771,451 (a national stage of PCT/US2019/056713, itself claiming a US provisional filed 17 October 2018) and US application 16/657,804, as incorporated by reference. The priority country is therefore the United States.

Why It Matters

A modern sequencing laboratory looks less like a bench and more like a distributed software system. A single patient’s tumor and normal samples might involve several instruments, multiple bioinformatics pipelines and separate teams, each doing its part on its own schedule. The generic computing pattern here is a microservice architecture, where many small independent programs each do one job and coordinate by sending messages.

That style of design has a known weakness: when several workers can pick up the same job, they sometimes do. The claims turn the cure, a broadcast that tells peers to stand down, into a defined step, together with a rule that tests which overlap are merged before any work starts. For a laboratory, duplicated sequencing or analysis costs real reagent and compute money and can delay a report, so a precise tracking layer is part of the product.

What the document does not do is claim a new sequencing method or a new biomarker. The listed assay steps are conventional ones, including next-generation sequencing, and the novelty argued is the orchestration. The scope therefore turns on the order-management arrangement in claim 1 rather than on any particular laboratory protocol.

Related Concepts

• Microservices – the architecture style in which each task is performed by a small independent service.
• Massively parallel sequencing – also called next-generation sequencing, the technique the final report is generated from.
• Microsatellite instability – a tumor marker the dependent claims score alongside the genetic variants.
• Tumor mutational burden – another score the dependent claims add to the report.
• Liquid biopsy – the blood-based, cell-free DNA specimen type covered by one dependent claim.

---

AU 2026202252 was published in the Australian Official Journal of Patents on 9 April 2026 and is open for public inspection. Patent applications represent inventions that are sought to be protected and do not necessarily reflect commercially available products.