Guides for diagnostic development

Designing a custom NGS panel: what to decide first

Decide what the result will be used for before you pick genes or chemistry. Intended use sets the variant types, specimen, sensitivity and validation level, and those decide almost everything else about the panel.

The decisions, in order

Six choices that shape the panel

1

Intended use and validation level

Research, GLP or clinical-trial use, a CLIA lab test, or an FDA submission. This sets how much validation you need and how locked-down the design must be.

2

Variant types

SNVs, indels, copy number, fusions, haplotypes, or signatures such as MSI and TMB. Each needs different design and analysis.

3

Specimen and input

FFPE, blood, cell-free DNA, saliva or another matrix, and how much DNA or RNA you can realistically get.

4

Sensitivity

Germline variants sit near 50% or 100% allele fraction; tumor and liquid biopsy variants may sit below 1%, needing deeper sequencing and unique molecular identifiers.

5

Content

Only the genes and regions the intended use needs. Every extra target adds cost, sequencing and validation work.

6

Chemistry, platform and throughput

Amplicon or hybrid capture, a sequencer that fits your volume and regulatory needs, and the turnaround your users expect.

Chemistry

Amplicon or hybrid capture?

Amplicon (PCR-based)

  • Best for small, focused panels: known hotspots and SNPs.
  • Works with low and degraded DNA.
  • Fast, simple workflow, often one day.
  • Very uniform coverage for small target sets.
  • Watch for allele dropout from primer-site variants; weak for CNVs and unknown fusion partners.

Hybrid capture

  • Best for larger panels, exomes, CNVs, fusions and novel variants.
  • Needs more input, though modern kits handle less.
  • Longer workflow with more steps.
  • More even coverage across large regions.
  • Higher cost per sample for small panels.

Validation strategy

Modularize the validation

Validate the workflow as three separate modules, each with defined inputs, outputs and acceptance criteria.

Pre-analytical

Extraction

Specimen types, nucleic acid yield and quality, input range and stability. Output: DNA or RNA that meets set specifications.

Analytical

Wet lab

Library preparation, amplification or capture, and sequencing. Output: data that meets set QC metrics such as coverage, uniformity and base quality.

Post-analytical

Bioinformatics

The pipeline from raw reads to reported variants or haplotypes, validated against fixed reference datasets and version-locked.

Because each module hands off to the next against a defined specification, a change to one module only needs that module revalidated. A new extraction kit, a new sequencer or a pipeline update can be revalidated on its own, with a small end-to-end confirmation, rather than repeating validation of the entire process. Set the handoff specifications up front, keep reference datasets for pipeline revalidation, and record every change under change control.

Quality control

Choose your controls

Pick controls during design, not after validation; the validation should prove they catch the failures you care about.

Batch positive control

Catches a run that misses known variants. Use well-characterized reference material or a cell line carrying each variant type on the panel, on every batch.

Sensitivity control

Catches loss of sensitivity near the limit. Use variants at about 1.5–2× the limit of detection, or the lowest allele fraction you report.

Negative / wild-type control

Catches false positives at targeted positions. Use material known to be negative for the panel's variants.

No-template control

Catches contamination in library prep and amplification. Carry water or buffer through from the first wet-lab step.

Extraction control

Catches extraction failure or contamination. Process it alongside specimens from extraction onward.

Per-sample QC

Catches a weak or failed sample. Set thresholds for coverage, uniformity, on-target rate and base quality, and check sample identity to catch swaps.

Match controls to your specimen matrix, such as FFPE reference material for an FFPE test or a cell-free DNA standard for liquid biopsy. Track control results over time to spot drift and test each new reagent lot against them. With a modular validation, each module's controls also show which module failed. Our GlutenID case study shows a custom NGS SNP panel and haplotype pipeline taken to FDA clearance.

Designing an NGS panel?

We design panels and bioinformatics pipelines, validate them module by module, and run them in our CLIA/CAP laboratory.