1 of 29

Innovation in Genetic Testing:�Designing and Curating for �Maximized Clinical Utility

Connolly Steigerwald, MS, CGC

June 3, 2026

2 of 29

DISCLOSURE

I am a full-time salaried employee of Ambry Genetics. Compensation includes performance-based incentives, and I own stock in Tempus AI, Inc.

3 of 29

Learning Objectives

Examine factors of test design and how they inform clinical utility

Summarize future directions and how they improve accuracy and yield of germline testing

Understand how long read sequencing and other technologies impact the diagnosis of genetic conditions

4 of 29

Demystifying Genetic Test Design

Genetic counselors play a vital role in patient care, yet may not have full visibility into developing genetic tests with true clinical value.

The term "the lab" is often used broadly, but behind it lies a complex, dynamic process that brings science to life for each patient report.

This presentation shines a light on what really happens inside a genetic diagnostic laboratory – helping GCs better understand the nuances, decision-making, and innovation that drive meaningful results for patients

5 of 29

How does test design �inform clinical utility?

6 of 29

What factors inform test design?

What do the clinicians want?

What does a payor agree to cover?

How do you make sure a filter doesn’t exclude a finding?

Can you validate a specific finding?

Which genes belong?

Before thinking about how test design informs clinical utility we have to understand test design itself

Test design is not a purely technical exercise; it is a multidimensional process that integrates wet-lab methodologies, bioinformatics pipelines, variant interpretation frameworks, and operational considerations. Each of these components directly influences clinical utility

What factors really inform test design, a few important ones:

Which genes belong: Is there gene disease validity between the finding and the phenotype and how strong is it?

Can you validate a specific finding: What are the common variant types in the gene, what are we aiming to find with the test we are doing vs. Technical limitation of assay (i.e. STR/SVs vs. short read capability)

Variant filtering: This where labs really begin to differentiate in their approach to analysis (e.g. difference in setting filter parameters, phenotype first approach vs. Genotype first approach)

Outside of that we have Payor coverage: What will get paid vs what won’t

What do clinicians want: Balancing act between different clinicians' needs and technical capability of assay.

Balance between what’s desired, whats possible, whats sustainable

7 of 29

Building a Test That Clinicians Want�Evidence Pipeline

Published Evidence

Consensus Guidelines

Test Design

Users

Payor Policy

8 of 29

How does test design inform clinical utility?

Assay + Analytics + Interpretation

“the test runs”

Test

ordered

Report in hand

9 of 29

How does test design inform clinical utility?

Assay + Analytics + Interpretation

Test

ordered

Report in hand

Wet Lab

Extractions

Library preparation

Primary assay

Secondary/orthogonal assays

Functional assays

Dry Lab

QC metrics

BI Pipelines

Variant calling algorithms

Data output

Data Assessment

Gene-disease relationships

Variant interpretation

Phenotype review

Summary of findings

10 of 29

Bioinformatic filtering and tools

Genetic data

Validation of variants

Variant classification, reclassification processes

Interpretation/reanalysis

Assessments to clarify the pathogenicity of variants RNA studies, structural assessment, family co-segregation

Orthogonal confirmation, product-specific cut offs

Annotation and filtering of sequencing data

Primary assay (NGS, Sanger), capture kit customizations, secondary assays

Report

Test ordered

Follow-up/ Functional studies

Clinical Utility of Genetic Testing

11 of 29

Future Directions: �Improving Diagnostic Potential

12 of 29

Even Tests With the Highest Clinical Utility Fall Short

No variants reported or missing second variant despite well-defined phenotype?

Phenotype only partially explained?

Variants of uncertain significance that may be the answer?

Over 50% of cases sent for clinical exome or genome sequencing fail to identify a diagnosis

13 of 29

Limitations of Current Short-Read Technology

Repetitive DNA sequences
Mobile or Transposable Elements, STRs, (micro/mini) Satellites, Telomeres
Segmental duplications
Genes with 1 or multiple pseudogenes with high sequence homology

Short read sequencing (SRS)

150-300 bp

14 of 29

The Multi-Omics Revolution: Seeing the Full Picture

Genomics (Long Reads)

Epigenomics

Transcriptomics

Proteomics

Metabolomics

15 of 29

The Multi-Omics Revolution: Seeing the Full Picture

Ali, S.S., Li, Q. & Agrawal, P.B. Implementation of multi-omics in diagnosis of pediatric rare diseases. Pediatr Res 97, 1337–1344 (2025). https://doi.org/10.1038/s41390-024-03728-w

16 of 29

Where Technical Advancement Meets Clinical Utility

17 of 29

A New Era of Genomics: �Long-Read Sequencing

20XX

18 of 29

The Potential of Long Read

Short read sequencing (SRS)

150-300 bp

Long read sequencing (LRS)

several kilobases and up to Mb scale

Benefits

Structural variants
Pseudogenes
Repeat expansions
Methylation status
Copy neutral variants/large copy number variants
Repeat-rich and high-GC content regions

Limitations

Best practices for analysis/interpretation are under development
Orthogonal confirmation may be a challenge
Lack of population frequency data for elusive variant types

The benefits of LRS are vast. It helps us accurately identify structural variants, navigate pseudogenes, detect repeat expansions, assess methylation status, and read copy neutral variants, large copy number variants, and repeat-rich or high-GC content regions.

This hummingbird example is everywhere but I found it so helpful when first understanding the differences so in the short read hummingbird puzzle, 1000 piece puzzle, a lot of the pieces are just background, not obvious where to put them so harder to put together

With long read, less, larger pieces of the puzzle, easier to see what’s not right

BUT still limitations

Population frequency data for previously elusive variants could pose interpretive challenges

Orthogonal confirmation is a big one as well, because we are now finding variatns previous technology wasn’t able to, so how are we making sure it’s truly there? Whats the best way to validate? And at scale?

19 of 29

LRS in the Clinic: Important Considerations

Operational Considerations

Sample requirements
Data long-term storage and reanalysis needs
Cost and reimbursement

Clinical Considerations

Patient consenting and resulting

Data Interpretation Considerations

Lack of reference databases for complex variants
Non-coding variants will require additional data to interpret
Methods for orthogonal confirmation

Potential for more uncertainty until ability to interpret catches up with technology

So as exciting as it is there are still limitations as this comes to clinic and as labs build this in, and we don’t want to overlook t hat

As with any piece of test design like we talked about earlier, Multiple perspectives to think about - operationally, clinically, and technically,

Operationally, we’re considering sample requirements and how this impacts need for long term storage and data

We have cost

In clinic, this might look like increased uncertainty in some settings until interpretation fully catches up to technology so we are yet to see that as well

And regarding interpretation – as we’ve discussed, we might be seeing a lot of new variants in fact that’s sort of the whole idea, so how are we going to make sure we have enough data to understand those

These are some of the important considerations when bringing long read to clinic and when critically thinking about “what is going into my genetic test”

Until we as a field can chip away at some of these limtations over time, naturally we will do our best at using other tools are our disposal to close diagnostic gaps

And one of those could be RNA

20 of 29

Exploring Transcriptomics:�The Power of RNA Analysis

20XX

21 of 29

Some Genetic Variants Have a �Negative Splicing Impact

INTRON

EXON 1

INTRON

EXON 2

INTRON

EXON 3

INTRON

EXON 4

DNA

EXON 5

MUTATION A

MUTATION B

MUTATION C

Missing Exons 3-5

Exon 3 Missed

Exon 4 Longer

RNA

PROTEIN

Truncated Protein

(Inactive)

Smaller Protein

(Inactive/Does not work properly)

Longer Protein

(Inactive/Does not work properly)

RNA Analysis

22 of 29

RNA Testing Provides Additional Insight

Establish pathogenicity for intronic VUS and missense variants with splice impact

Increase confidence in likely pathogenic and pathogenic classifications of intronic variants

Correct misclassification of canonical variants where observed impact ≠ predicted impact

Uncover clinically significant synonymous variants

23 of 29

Depicting Change in Variant Classifications Based on RNA Evidence

305 VUS downgrades
Potentially clinically actionable upgrades were made in 97 cases

70 VUS reclassified to P/LP
27 deep intronic variants that were previously unreported

Genes included: APC, ATM, BRCA1,BRCA2, BRIP1, CDH1, CHEK2, MLH1,MSH2, MSH6, MUTYH, NF1, PALB2,PMS2, PTEN, RAD51C, RAD51D, andTP53

Horton C, Hoang L, Zimmermann H, Young C, Grzybowski J, Durda K, Vuong H, Burks D, Cass A, LaDuca H, Richardson ME, Harrison S, Chao EC, Karam R. Diagnostic Outcomes of Concurrent DNA and RNA Sequencing in Individuals Undergoing Hereditary Cancer Testing. JAMA Oncol. 2024 Feb 1;10(2):212-219. doi: 10.1001/jamaoncol.2023.5586. PMID: 37924330; PMCID: PMC10625669.

24 of 29

Reclassification Rate in Neurology Patients Who Underwent RNA Analysis

Based on splicing events and impacts observed in RNA, 79% (30/38) of the variants changed classification

Adapted from Ichikawa S, Yergert K, Gasser B, Towne M, Burow D. Utility of Targeted RNA Analysis in Neurogenetic Disorders. Neurol Genet. Published online February 20, 2026. doi:10.1212/NXG.0000000000200341. Changes include image cropping and summarization of findings.

25 of 29

The Real-World Value of �Technical Advancements

26 of 29

16 year old, daughter of consanguineous parents
Dx at 4 yo with neuronal ceroid lipofuscinosis type 2 (CLN2) based on TPP1 enzyme analysis
Autosomal recessive lysosomal storage disorder

Build-up of ceroid lipofuscin and ultimately neuronal injury and death
Caused by variants in the TPP1 gene

26

Absent TPP1 enzyme activity

Normal clinical genetic testing

Confirmed Phenotype, Elusive Genotype

This is a patient example actually from my time in clinic

We saw a 16 year old female born to consanguinous parents

At 4 yo she had been diagnosed with neuronal ceroid lipofuscinosis type 2 also known as CLN2 or batten disease

She was diagnosed based on tpp1 enzyme analysis which by the age of 16 had become quite progressed

CLN2 disease is an autosomal recessive Lysosomal storage disorder leading to build-up of ceroid lipofuscin, and ultimately neuronal injury and death, caused by variants in the TPP1 gene. Typical diagnostic approaches have included both genetic testing of TPP1 or looking at the enzyme level of TPP1 bc

Individuals typically have absent TPP1 enzyme activity.

At 10 years old (around 2015) she had TPP1 sequencing including 20 bases of non coding sequence, an array, and again had sequencing of TPP1 in 2021 at a different lab all of which were negative

It was a challenge at the time because the diagnosis was rather certain, we had the enzyme analysis, the phenotype, and the family history, but no genetic diagnosis

27 of 29

27

Absent TPP1 enzyme activity

Normal clinical genetic testing

Younger sibling with absent TPP1 activity and normal clinical genetic testing

Targeted LRS homozygous deep intronic variant (c.1146-199 G>A) in intron 9 of TPP1

Heterozygous in parents

Confirmed Phenotype, Elusive Genotype

We then saw this probands sibling

She presented as a healthy 4 week old given this family history

She had had prenatal enzyme analysis of TPP1, given the inability to do genetic testing, and it was absent

Enzyme testing on dried blood spot done after birth confirmed absent TPP1 activity

She also had sequencing of TPP1 on buccal swab

Unsurprisingly no variants were identified

From here, the proband and her parents were enrolled in a targeted long read sequencing research study meant to identify variants missed by standard clinical testing at that time in 2023

Testing revealed a homozygous deep intronic variant (c.1146-199 G>A) in intron 9 of TPP1, which was predicted by SpliceAI to create a novel splice acceptor site

Subsequent targeted clinical testing confirmed the presence of this homozygous variant in the proband and younger sibling and found that the parents were heterozygous carriers of the variant

28 of 29

RNA studies (blood)

Proband and parents - altered RNA splicing
Parents – both wild type and aberrant splice products

Aberrant splice product 🡪 stop codon 🡪 nonsense-mediated decay

Upgraded to likely pathogenic per ACMG/AMP criteria

The Transcriptome Tells the Tale

We can’t forget about RNA though. RNA studies in the blood from proband and parents demonstrated altered RNA splicing

So there was an aberrant splice product with retention of 84 nucleotides of intron 9 between exons 9 and 10

This abnormal splice product was predicted to result in a stop codon and ultimately lead to nonsense-mediated decay of the mRNA transcript or premature protein truncation

In blood samples from parents, both wild-type and aberrant splice products were detected

The variant was subsequently upgraded to likely pathogenic per ACMG/AMP criteria.

This was an impactful case for me because we really needed each piece of the puzzle to reach an answer. Needed that enzyme analysis and family history to tip us off, needed long read to findf the variant itself and needed RNA to tell us what that variant was doing. The diagnosis itself was not a mystery here, we knew what we were searching for, but we needed the right tools to do so

�So hopefully that was an informative illustration of how important use of emerging technologies can be especially when used together to close some of these diagnostic gaps that persist in the field.

I also hope going forward, as you’re ordering genetic testing for your patients and working with clinical testing labs, you have better context and transparency into what’s going on behind the scenes in test design �

29 of 29

Thank You!

�csteigerwald@ambrygen.com