Custom Targeted Sequencing Panels: Designing and Validating Panels for Specific Gene Sets and Genomic Regions

Off-the-shelf sequencing panels — whether a 50-gene cancer hotspot kit or a comprehensive exome capture platform — serve the majority of clinical and research applications well, but they leave a substantial fraction of use cases unaddressed. A pediatric cardiologist needs a 30-gene panel focused exclusively on inherited arrhythmia and cardiomyopathy genes, not a 500-gene pan-cancer panel. A pharmacogenomics laboratory serving a diverse population requires allele coverage tuned to ancestry-specific variant frequencies that commercial European-centric panels systematically underrepresent. A research group fine-mapping GWAS loci needs probes targeting non-coding regulatory intervals that fall entirely outside the capture space of every commercial exome kit. In each of these scenarios, a custom targeted sequencing panel — designed, synthesized, and validated for the specific gene set or genomic region of interest — is the appropriate tool. This article examines when custom panels are warranted, how they are designed and validated, what enrichment chemistry and bioinformatic considerations govern their performance, and what costs and timelines researchers and clinicians should anticipate.

Figure 1: Custom Panel Decision Framework — When Off-the-Shelf Panels Fall Short

When Off-the-Shelf Panels Are Not Enough

The decision to commission a custom panel rather than purchase a commercial kit turns on four questions. First, does the required gene set align with any existing commercial panel? A cancer predisposition panel of 30 to 40 genes — ATM, BRCA1, BRCA2, CHEK2, PALB2, TP53, and the Lynch syndrome genes — is well served by multiple commercial offerings. A panel targeting 18 genes implicated in a specific rare disease pathway likely is not. Second, does the commercial panel cover the relevant variant types? Many commercial panels are validated for single-nucleotide variants and small insertions and deletions but provide inadequate coverage of copy-number variants, structural rearrangements, or repeat expansions — variant classes that may be central to the disease biology of interest. Third, does the commercial panel represent the populations under study? A 2025 pharmacogenomics panel study at the University of Florida (Lteif et al., Clinical and Translational Science) developed GatorPGx Plus specifically because commercial PGx panels underrepresented alleles common in African and Hispanic populations — CYP2D6 *17 and *29, CYP2C19 *9 — that direct clinical prescribing decisions. The panel's 62 variants across 14 genes, selected by requiring allele frequency above approximately 1 percent in any major ancestral population, produced actionable pharmacogenomic findings in 99 percent of participants in a diverse cohort. Fourth, does the panel need to cover non-coding regions? GWAS fine-mapping projects, enhancer characterization studies, and investigations of deep intronic splicing variants all require capture of genomic intervals that commercial coding-region panels do not target.

The three dominant application domains for custom panels are clinical oncology, inherited disease diagnostics, and pharmacogenomics. In oncology, custom panels enable integration of institution-specific biomarker priorities, inclusion of genes relevant to active clinical trials, and coverage of intronic regions harboring recurrent fusion breakpoints. In inherited disease, custom panels allow laboratories to curate gene lists based on the latest gene-disease validity evidence — removing genes with disputed associations that generate unwanted secondary findings and adding newly discovered disease genes that commercial panels, with their multi-year update cycles, have not yet incorporated. A 2025 study of virtual gene panels for inherited rare diseases (Sheikh Hassani et al., Clinical Chemistry) demonstrated that static panels missed disease-causing variants in 3.5 percent of cases due to gene list staleness, whereas flexible virtual panels with reflex expansion capabilities captured these diagnoses. In pharmacogenomics, custom panels can be designed for the specific drugs in a health-system formulary, avoiding the cost and interpretive burden of reporting variants irrelevant to local prescribing practice.

Figure 2: Custom Panel Design and Validation Pipeline — From Gene Selection to Clinical Report

Panel Design — From Gene Selection to Probe Synthesis

The design of a custom targeted panel proceeds through four stages: gene and region selection, probe design and in silico quality control, probe synthesis, and wet-lab analytical validation. Each stage involves decisions that affect the panel's clinical sensitivity, specificity, and long-term maintainability.

Gene and region selection begins with a systematic review of gene-disease validity evidence using frameworks such as ClinGen, which classifies gene-disease associations as definitive, strong, moderate, or limited. For diagnostic panels, only genes with definitive or strong evidence should be included; including genes with limited evidence inflates the variant-of-uncertain-significance rate without improving diagnostic yield. For research panels, the threshold may be relaxed to moderate or even limited evidence when the goal is discovery rather than clinical reporting. Target regions are then defined: for each selected gene, all coding exons plus a minimum of 10 to 25 base pairs of flanking intronic sequence are included to capture canonical splice sites. Known deep intronic pathogenic variants, promoter mutations, and untranslated region variants with functional evidence can be added as custom intervals. For panels targeting non-coding regulatory regions, intervals are defined by statistical fine-mapping (using tools such as FINEMAP or SuSiE) combined with epigenomic annotation from tissue-matched datasets (H3K27ac, ATAC-seq, DNase-seq).

Probe design converts the target interval list into a set of oligonucleotide sequences optimized for capture uniformity and specificity. The target territory size dictates the enrichment strategy: for panels below approximately 50 to 200 kilobases (roughly 50 to 500 genes depending on gene size), amplicon-based enrichment using multiplex PCR is the faster and less expensive option, requiring as little as 1 to 10 nanograms of input DNA and completing library preparation in a single workday. For panels exceeding approximately 200 kilobases, or for panels requiring fusion detection, structural variant calling, or coverage of GC-extreme regions, hybrid capture with biotinylated oligonucleotide probes is preferred. Probe design parameters include probe length (typically 80 to 120 base pairs, with longer probes tolerating more sequence variation), tiling density (1.5-fold to 2-fold coverage of each target base), and GC content balancing (probes with GC content below 25 percent or above 75 percent are flagged for compensatory density increases or alternate-strand redesign). Repeat masking and pseudogene cross-reactivity screening — in which candidate probes are BLASTed against the reference genome to identify those with significant secondary alignments — eliminate probes that would capture off-target sequences and reduce effective on-target depth.

A 2025 study from Tan Tock Seng Hospital in Singapore (Das et al., Scientific Reports) exemplifies the modern custom panel design-and-validation workflow. The TTSH-oncopanel, a 61-gene hybrid-capture panel targeting clinically actionable mutations across solid tumor types, was designed by curating genes with tier-I evidence for targeted therapy, prognosis, or clinical trial eligibility. The panel was validated on 43 clinical specimens and external quality assessment samples, achieving 98.23 percent sensitivity, 99.99 percent specificity, and 99.99 percent accuracy for single-nucleotide variants and small insertions and deletions, with a limit of detection below 5 percent variant allele frequency. The in-house panel reduced turnaround time from approximately three weeks — the typical interval when samples are sent to an external reference laboratory — to four days, a clinically meaningful acceleration for patients awaiting genomically guided therapy decisions.

Figure 3: Enrichment Chemistry Comparison — Hybrid Capture vs. Amplicon-Based Panel Performance

Library Preparation and Sequencing — Choosing the Right Enrichment Chemistry

The choice between hybrid capture and amplicon-based enrichment is the most consequential technical decision in panel design, with direct implications for variant detection sensitivity, workflow complexity, and per-sample cost. A 2025 study of 1,211 non-small-cell lung cancer specimens at Weill Cornell Medicine (Dillard et al., Journal of Molecular Diagnostics) quantified the clinical impact of this choice. The standard workflow used an amplicon-based DNA and RNA sequencing panel (Oncomine) as the first-line test, with negative cases reflexed to a hybridization-capture RNA sequencing panel (TruSight Oncology 500). Among 120 cases that were driver-negative by amplicon testing, the hybridization-capture reflex identified nine additional clinically actionable gene fusions — involving ALK, BRAF, NRG1, NTRK3, ROS1, and RET — that the amplicon panel had completely missed. An analysis of the AACR Project GENIE database (version 15.1, 20,900 NSCLC cases) suggested that approximately 17.4 percent of clinically relevant fusions would be missed by amplicon-only testing, primarily because amplicon designs require prior knowledge of both fusion partners and breakpoints, whereas hybrid capture probes tiling across introns can capture rearrangements with unknown partners.

For applications other than fusion detection, the amplicon-versus-capture trade-offs are well characterized: amplicon methods offer faster turnaround, lower input requirements, and higher on-target rates (typically above 90 percent), but suffer from primer-binding site dropout in the presence of variants under primer sequences and reduced uniformity in GC-rich or GC-poor regions. Hybrid capture provides more uniform coverage, better tolerance for degraded DNA (important for formalin-fixed paraffin-embedded specimens), and the ability to scale to large target territories, but requires higher DNA input (typically 50 to 200 nanograms) and an overnight hybridization step that extends the workflow by approximately 16 hours.

Sequencing depth is application-dependent. For inherited disease panels where heterozygous germline variants are the target, 100-fold to 300-fold mean coverage is standard, with a minimum of 20-fold to 30-fold coverage at every target base. For oncology panels requiring detection of somatic variants at 5 percent allele fraction, 500-fold to 750-fold coverage is recommended. For liquid biopsy panels targeting circulating tumor DNA at variant allele fractions below 1 percent, depths of 1,000-fold to 5,000-fold may be necessary, typically achieved by reducing panel size to focus sequencing resources on the highest-value targets. A 2025 validation of a 1,021-gene pan-cancer panel (Meintani et al., International Journal of Molecular Sciences) across over 1,300 solid tumor patients demonstrated that at a mean depth exceeding 500-fold, the panel detected on-label treatment biomarkers in 12.57 percent of patients — rising to 20.15 percent when immunotherapy biomarkers (tumor mutational burden and microsatellite instability) were included — with 70 percent concordance between matched formalin-fixed tissue and plasma circulating tumor DNA.

Figure 4: Bioinformatics for Custom Panel Data — From Aligned Reads to Annotated Variants and CNV Calls

Bioinformatics for Custom Panels — Variant Calling, CNV Detection, and Custom Annotation

The bioinformatic analysis of custom panel data shares its core workflow with exome and genome analysis — read alignment with BWA-MEM, duplicate marking, base quality score recalibration, and variant calling with GATK HaplotypeCaller or similar tools — but introduces several requirements specific to targeted panels. The most important is copy-number variant detection, which is both more challenging and more clinically important in panel data than in exome data, because panels are frequently used in oncology and inherited disease settings where gene-level deletions and duplications are common pathogenic mechanisms.

A comprehensive 2025 benchmarking study (Munté et al., Briefings in Bioinformatics) evaluated 12 CNV detection tools on four validated gene panel datasets, assessing 107 individual tool parameters and 66 tool-pair meta-caller combinations. The top-performing individual tools were ClinCNV and GATK-gCNV, with GATK-gCNV offering the highest sensitivity. Meta-calling — combining the output of two or more individual callers — improved precision at a modest cost to sensitivity and is recommended for clinical applications where false-positive CNV calls trigger unnecessary follow-up testing. The study identified 13 parameter values that consistently improved F1 scores across tools, including adjustments to transition probability parameters in hidden Markov model-based callers and minimum exon count thresholds for calling.

Critical quality-control steps for panel-based CNV analysis include construction of a panel of normals — a set of at least 10 to 30 samples processed with the same library preparation and sequencing protocol, against which test-sample coverage profiles are normalized — and loess-based GC bias correction that accounts for the relationship between local GC content and capture efficiency. For single-exon CNV calls, orthogonal confirmation by multiplex ligation-dependent probe amplification (MLPA) or quantitative PCR is standard practice, as single-exon events have elevated false-positive rates even with optimized calling algorithms.

Custom annotation pipelines are required when panels target genes or regions not covered by standard annotation resources. For pharmacogenomics panels, star-allele calling — inferring haplotypes from combinations of individual variants — requires specialized tools such as PharmCAT or Stargazer that interpret variant phase and translate variant combinations into predicted metabolizer phenotypes. A 2026 validation of a 335-gene pharmacogenomics panel (Ramudo-Cela et al., Pharmaceuticals) demonstrated that custom annotation pipelines integrating star-allele calling with copy-number analysis achieved 98 percent concordance with GeT-RM reference haplotypes and successfully resolved complex CYP2D6 structural variants including hybrid genes and multi-copy alleles that are invisible to standard variant callers.

Figure 5: Cost and Timeline Breakdown for Custom Panel Projects

Cost and Timeline — From Design to Routine Testing

The cost of a custom targeted sequencing panel has three components: design and synthesis, validation, and per-sample sequencing. Design and probe synthesis costs for a hybrid-capture panel typically range from $3,000 to $15,000 depending on target territory size, probe count, and the complexity of the repeat-masking and pseudogene-discrimination analysis required. Amplicon panels, which use synthesized primer pools rather than biotinylated probes, are generally less expensive to design and synthesize — roughly $1,500 to $5,000 for panels of 50 to 500 amplicons — but incur higher per-sample costs at scale due to the primer synthesis and pooling overhead.

Analytical validation — establishing the panel's sensitivity, specificity, reproducibility, and limit of detection — typically requires 30 to 50 well-characterized reference samples and costs $10,000 to $30,000 in reagent and labor expenses. Validation should assess the panel's performance on the sample types that will be used in production (blood, saliva, fresh-frozen tissue, FFPE, or plasma), at the input amounts specified in the protocol, and should include samples with known pathogenic variants spanning the variant types the panel is designed to detect: single-nucleotide variants, insertions and deletions, copy-number variants, and, where applicable, gene fusions and microsatellite instability.

Per-sample sequencing costs depend on panel size, target depth, and sample multiplexing. For a 500-kilobase hybrid-capture panel sequenced at 500-fold mean depth on an Illumina NovaSeq X Plus, a single lane can accommodate approximately 48 to 96 samples depending on the target read count per sample, yielding a per-sample sequencing cost of approximately $50 to $120 for reagents alone. At these economics, the per-sample cost of a custom panel is comparable to or slightly lower than a commercial panel of equivalent size, with the custom panel offering the advantage of a precisely tailored gene set.

The total timeline from project initiation to routine testing is typically 10 to 16 weeks: 2 to 4 weeks for gene selection and target interval definition, 4 to 6 weeks for probe design, synthesis, and quality control, 3 to 4 weeks for analytical validation on reference samples, and 1 to 2 weeks for bioinformatic pipeline validation and reporting template development. Expedited timelines of 6 to 8 weeks are feasible when gene lists are pre-curated and the probe synthesis vendor offers fast-track manufacturing. For laboratories that anticipate iterative panel updates — adding newly discovered disease genes, removing genes with downgraded evidence, or expanding coverage to additional non-coding regions — establishing a relationship with a probe synthesis vendor that supports small-batch re-synthesis and panel versioning is advisable from the outset.

FAQ

When should I design a custom targeted panel rather than using a commercial kit?

Design a custom panel when the required gene set is not covered by any existing commercial panel, when the commercial panel does not detect the variant types relevant to your application (for example, copy-number variants or gene fusions), when the populations you serve are underrepresented in the commercial panel's allele frequency databases, or when your application requires targeting of non-coding genomic regions — GWAS loci, enhancers, deep intronic splice sites — that fall outside commercial coding-region capture space.

What is the difference between hybrid capture and amplicon-based targeted sequencing?

Hybrid capture uses biotinylated oligonucleotide probes to enrich target regions from a sequencing library and supports large panels of hundreds to thousands of genes with uniform coverage and fusion detection capability. Amplicon-based methods use multiplex PCR for direct target amplification and offer faster turnaround and lower DNA input requirements but are limited to smaller panels, are vulnerable to primer-site variants causing allele dropout, and cannot detect fusions with unknown partner genes. A 2025 clinical study found that amplicon-based testing missed approximately 17 percent of actionable gene fusions in lung cancer that were subsequently identified by hybrid-capture reflex testing.

What sequencing depth do I need for a custom targeted panel?

For inherited disease panels detecting heterozygous germline variants, 100-fold to 300-fold mean coverage with a minimum of 20-fold to 30-fold at every target base is standard. For oncology panels detecting somatic variants at 5 percent allele fraction, 500-fold to 750-fold is recommended. For liquid biopsy applications detecting circulating tumor DNA below 1 percent variant allele frequency, 1,000-fold to 5,000-fold depth is required.

How are copy-number variants detected from targeted panel data?

Panel-based CNV detection uses read-depth comparisons between test samples and a panel of normals — a set of reference samples processed identically — to identify regions with statistically significant deviations in normalized coverage. The leading tools as of 2025 are ClinCNV and GATK-gCNV, with meta-calling strategies combining multiple algorithms recommended for clinical applications. Single-exon CNV calls require orthogonal confirmation by MLPA or quantitative PCR.

How long does it take to develop and validate a custom targeted panel?

A typical timeline is 10 to 16 weeks from project initiation to routine clinical or research testing: 2 to 4 weeks for gene and interval selection, 4 to 6 weeks for probe design and synthesis, 3 to 4 weeks for analytical validation on reference samples, and 1 to 2 weeks for bioinformatic pipeline validation. Expedited timelines of 6 to 8 weeks are feasible when gene lists are pre-curated and the synthesis vendor offers fast-track manufacturing.

What does a custom targeted panel cost?

Design and probe synthesis costs range from $3,000 to $15,000 for hybrid-capture panels and $1,500 to $5,000 for smaller amplicon panels. Analytical validation typically requires $10,000 to $30,000 in reagent and labor costs. Per-sample sequencing costs at 500-fold depth range from approximately $50 to $120 depending on panel size and multiplexing strategy. The total upfront investment of $15,000 to $45,000 is amortized across the sample cohort, making custom panels cost-effective for studies exceeding approximately 100 samples.

References:

1. Das K, Tay MLI, Yong EY, Chuah KL. A targeted next-generation sequencing panel for identification of clinically relevant mutation profiles in solid tumours. Scientific Reports. 2025;15:20740. https://doi.org/10.1038/s41598-025-08039-6
2. Dillard A, Xu K, Sun Y, Lin HH, Shen C, Song E, Saxena A, Hissong E, Yemelyanova A, Lindeman NI, Velu PD, Solomon JP. Comparison of targeted RNA-sequencing platforms for oncogenic fusion detection in non-small-cell lung cancer. Journal of Molecular Diagnostics. 2025;27(6):438-445. https://doi.org/10.1016/j.jmoldx.2025.02.007
3. Munté E, Roca C, Del Valle J, Feliubadaló L, Pineda M, Gel B, Castellanos E, Rivera B, Cordero D, Moreno V, Lázaro C, Moreno-Cabrera JM. Detection of germline CNVs from gene panel data: benchmarking the state of the art. Briefings in Bioinformatics. 2025;26(1):bbae645. https://doi.org/10.1093/bib/bbae645
4. Meintani A, Ozdogan M, Touroutoglou N, Papazisis K, Boukovinas I, Bilir C, et al. Comprehensive evaluation of a 1021-gene panel in FFPE and liquid biopsy for analytical and clinical use. International Journal of Molecular Sciences. 2025;26(13):5930. https://doi.org/10.3390/ijms26135930
5. Lteif C, Gawronski BE, Cicali EJ, Martinez KA, Newsom KJ, Starostik P, Cavallari LH, Duarte JD. Development of an ancestrally inclusive preemptive pharmacogenetic testing panel. Clinical and Translational Science. 2025;18(5):e70230. https://doi.org/10.1111/cts.70230
6. Sheikh Hassani M, Jain R, Ramaswamy V, et al. Virtual gene panels have a superior diagnostic yield for inherited rare diseases relative to static panels. Clinical Chemistry. 2025;71(1):169-184. https://doi.org/10.1093/clinchem/hvae183
7. Ramudo-Cela L, Izquierdo-Garcia M, Dolores-Sequedo M, Cubells-Perez V, Bernal S, Riera P, et al. Analytical and clinical validation of Action PharmaKitDx: a comprehensive NGS panel for the identification of pharmacogenetic variants in diverse populations. Pharmaceuticals. 2026;19(4):568. https://doi.org/10.3390/ph19040568

For research purposes only, not intended for clinical diagnosis, treatment, or individual health assessments.