Built for clinical oncology labs that deliver therapy-matched reports, not just a variant list. AMP-guided somatic classification, TMB and MSI assessment, and complete tumor profiling in one platform.
Modern oncology testing has outgrown simple variant detection. Clinical labs now deliver therapy-matched reports that connect a patient's tumor profile to FDA-approved treatments, ongoing clinical trials, and resistance mechanisms, within a turnaround time that fits the clinical workflow.
This work depends on a specific set of capabilities: distinguishing oncogenic somatic drivers from germline background, classifying variants against the AMP/ASCO/CAP evidence tiers, integrating biomarkers like Tumor Mutational Burden and Microsatellite Instability, and generating reports oncologists can act on directly. Each step builds on the previous one, so a disconnected toolchain of separate variant callers, manual database lookups, and templated report editors compounds delays and interpretation risk.
Golden Helix solves this as a production system. VarSeq and VSClinical AMP run annotation, filtering, classification, and reporting in a single deterministic workflow, with every interpretation decision traceable to its evidence source. For the bioinformatic principles behind the workflow, see the somatic variant analysis guide.
Most labs evaluating precision oncology software focus on variant-calling sensitivity. That is table stakes. The differentiators emerge in the interpretation layer: the decisions a platform makes, or fails to make, after variants are called. These five requirements separate clinical-grade software from research tools adapted for clinical use.
The joint AMP/ASCO/CAP consensus guidelines (Li et al., J Mol Diagn, 2017) define a four-tier system for classifying somatic variants by clinical actionability. Tier I variants carry strong clinical significance with FDA-approved therapies or professional guidelines. Tier IV variants carry benign or likely benign evidence. The tiers in between require structured evidence review that, done manually, introduces variability across interpreters and samples.
Clinical-grade software scores the criteria underpinning each tier from hotspot databases, population frequency filters, functional evidence, and therapeutic associations, then lets a qualified pathologist review and override the pre-classification before sign-out. Evaluate for automated tier pre-classification, an auditable evidence trail per criterion, and a clear separation between auto-classification and final pathologist sign-off.
Comprehensive cancer panels like Illumina TSO-500 detect the full spectrum of somatic alterations: small variants (SNVs and indels), copy number variants (CNVs), structural variants and gene fusions, and genomic signatures (TMB, MSI, HRD). A platform that handles SNVs robustly but delegates CNV and fusion interpretation to a separate tool creates reporting gaps and workflow friction.
Evaluate whether the software supports integrated somatic CNV detection with LOH analysis, gene fusion and structural variant detection with domain and exon boundary evaluation, and direct TMB and MSI calculation from NGS data, all in the same sample session.
Variant classification without therapeutic context produces incomplete reports. Oncology labs need curated evidence linking specific variants to FDA-approved therapies, EMA approvals, NCCN guidelines, and relevant clinical trials, surfaced at the moment of interpretation rather than as a separate research step.
Aggregated databases like CIViC and COSMIC provide community evidence but require the interpreting scientist to synthesize relevance. A professionally curated knowledgebase like Golden Helix CancerKB™ provides pre-synthesized, tumor-type-specific interpretations with report-ready language, cutting per-sample interpretation time.
Matched tumor-normal analysis is the clinical gold standard for somatic calling: the matched normal sample filters out germline variants and sequencing artifacts, leaving a cleaner somatic call set. A significant proportion of clinical cases still arrive as tumor-only samples, including liquid biopsies, archival FFPE tissue, and cases where normal tissue is unavailable or cost-prohibitive.
Tumor-only analysis is technically harder. Without a matched normal, the platform uses population frequency databases (gnomAD, ExAC) and variant-type heuristics to distinguish likely somatic mutations from germline variants in the tumor sample. Evaluate whether the software supports both modes and whether the tumor-only filtering strategy is configurable and auditable.
Many solid tumor panels now detect both somatic drivers and hereditary cancer risk variants (BRCA1/2, Lynch syndrome genes, TP53) in a single test. This combined workflow requires a platform that applies AMP classification criteria to somatic findings and ACMG classification criteria to germline findings within the same sample session, with appropriate segregation in the final report. Many platforms handle this poorly, either requiring two separate analysis runs or conflating somatic and germline evidence in the classification step. Confirm that the software supports combined interpretation with distinct evidence frameworks for each finding type. For the germline side, review VarSeq clinical interpretation capabilities.
Evidence Levels
Standardized AMP evidence ranking from FDA-approved through pre-clinical studies.
Integrated genomic signatures for immunotherapy response prediction.
Expert-curated clinical interpretations and therapy associations.
VSClinical AMP surfaces therapeutic context as part of the interpretation, so interpreting scientists do not leave the workflow to research relevance. Golden Helix CancerKB™ provides the curated layer alongside community databases like CIViC and COSMIC.
Connect somatic mutations to FDA, EMA, and NCCN-recognized therapies with drug descriptions and resistance profiles.
Match patients to ongoing trials from ClinicalTrials.gov based on tumor profile, location, and phase requirements.
A KRAS G12C variant in lung adenocarcinoma reads differently than the same variant in colorectal cancer. CancerKB™ reflects that context explicitly.
A complete oncology report requires the full spectrum of somatic alteration. VarSeq integrates small variants, structural variations, and genomic signatures in one sample session.
Identify amplification and deletion events that drive oncogenesis. Integrated LOH analysis distinguishes copy-neutral events and somatic-only changes.
Somatic CNV DetectionDetect structural variants and gene fusions like ALK, BCR-ABL1, and EGFR exon 19/21. Evaluate domain and exon boundaries and tier evidence across tumor types.
CNV & SV AnalysisCalculate Tumor Mutational Burden and Microsatellite Instability directly from NGS data, with configurable thresholds for immune checkpoint inhibitor reporting.
All five requirements are met inside one integrated workflow. Three components work together to take a sample from variant call to oncologist-ready report.
| Component | Role in Oncology Workflow | Key Capabilities |
|---|---|---|
| Sentieon | Secondary analysis: alignment and somatic variant calling | TNscope (tumor-normal), TNhaplotyper2 (tumor-only), FFPE artifact handling, structural variant calling |
| VarSeq + VSClinical AMP | Tertiary analysis: annotation, filtering, classification, reporting | AMP auto-classification, CancerKB™ integration, CNV and fusion analysis, TMB/MSI calculation, combined somatic-germline support |
| VSWarehouse | Enterprise variant data management | Somatic catalog storage, AMP classification history, cohort-level biomarker analysis, audit trail |
Sentieon provides the somatic variant-calling foundation. TNscope handles matched tumor-normal analysis, with sensitivity calibrated for low-allele-frequency somatic variants (down to 1 to 5% VAF) common in heterogeneous tumors and liquid biopsies. TNhaplotyper2 handles tumor-only cases, applying population frequency modeling to distinguish likely somatic calls from germline background without a matched normal.
Sentieon is validated against GATK4 and produces deterministic results: the same input produces the same output every run, which matters for CLIA validation and reproducibility audits. Learn more about Sentieon.
VarSeq handles the annotation, filtering, and clinical classification layer that turns a raw VCF into an interpreted variant list. For oncology, that means:
As oncology programs grow, variant data accumulates into a strategic asset when it stays accessible. VSWarehouse stores somatic variant projects and AMP classification catalogs in a structured database that supports cohort-level queries, longitudinal sample tracking, and classification consistency review. Labs running high-volume panels can query variant history across samples to accelerate interpretation of recurrent hotspot variants and identify cohort-level biomarker trends. Learn more about VSWarehouse.
Golden Helix supports the full spectrum of clinical oncology testing configurations in use today, from matched tumor-normal panels to ctDNA liquid biopsy.
Matched tumor and normal samples analyzed together. TNscope calls somatic variants with germline subtraction; VarSeq annotates and VSClinical AMP classifies. Recommended for solid tumor panels where matched normal tissue is available.
Tumor sample analyzed without a matched normal. TNhaplotyper2 applies population frequency filtering to reduce germline contamination. Appropriate for liquid biopsies, FFPE cases, and programs where matched normals are operationally unavailable.
A single test delivers somatic driver findings and hereditary cancer risk findings (BRCA1/2, MLH1/MSH2/MSH6/PMS2, TP53, and others), with AMP and ACMG classification applied to their respective finding types.
Pre-configured workflow for TSO-500 comprehensive genomic profiling data, including TMB, MSI, and fusion detection in the same analysis session.
Sensitivity calibrated for ctDNA variants at low allele frequencies. Appropriate for treatment monitoring, minimal residual disease (MRD) assessment, and cases where tissue biopsy is unavailable.
Somatic mutations may occur in as few as 1 to 5% of sequence reads due to tumor heterogeneity and biopsy composition. New cancer-gene associations, targeted therapies, and clinical trials publish continuously, so a Tier II classification today may be Tier I in six months as evidence accumulates.
Clinical-grade software must be accurate at the point of interpretation and structured to support reclassification as evidence evolves. The bioinformatic principles are covered in the somatic variant analysis resource.
Clinical oncology labs operate under strict regulatory requirements. Golden Helix supports on-premises, private cloud, and air-gapped deployment so labs can match their infrastructure and data governance requirements. For full lab infrastructure context, see the clinical lab infrastructure guide.
Full deployment within the institution's own infrastructure. No patient data leaves the institutional boundary. Complete data sovereignty for sites with strict security policies against cloud-based PHI storage.
Deployment within the lab's own AWS or Azure environment. Full administrative control, geographic data residency selection, and elastic compute scaling for variable sample volumes.
Fully offline deployment for facilities with strict network isolation requirements. All software, annotation databases, and licensing operate on an isolated internal network.
Reproducibility and audit. For CLIA-validated workflows, VarSeq produces deterministic output, essential for reproducibility documentation. VSWarehouse maintains a complete audit trail of variant classifications, evidence sources, and pathologist actions for each sample. Golden Helix operates under an ISO 13485-certified Quality Management System, and VarSeq Dx is CE marked under IVDR 2017/746. Review security and compliance capabilities.
Precision oncology software: what makes it clinical-grade, how AMP classification works, and how it fits your lab workflow.
Clinical-grade precision oncology software is designed for regulated diagnostic use in a CLIA-certified environment. The key distinctions are determinism (the same input produces the same output every run, enabling reproducibility validation), an auditable evidence trail for every classification decision, separation between automated pre-classification and pathologist sign-off, and ongoing curation of the underlying knowledge bases.
Research-grade tools are often optimized for discovery workflows, favoring sensitivity over specificity and flexibility over reproducibility, which makes them poorly suited to clinical production. Golden Helix VarSeq and VSClinical AMP are built for the clinical production context: validated workflows, deterministic pipelines, and a complete audit trail from variant call to signed report.
Both modes are supported. Matched tumor-normal analysis using Sentieon TNscope is the recommended approach when a matched normal is available, since germline subtraction produces a cleaner somatic call set. For tumor-only cases such as liquid biopsies, archival FFPE tissue, or programs where matched normals are operationally unavailable, Sentieon TNhaplotyper2 applies population frequency filtering using gnomAD and ExAC to distinguish likely somatic mutations from germline variants.
Both modes integrate directly into VarSeq for annotation and VSClinical AMP for classification without a workflow change.
VSClinical AMP scores each variant against the evidence criteria defined in the joint AMP/ASCO/CAP guidelines (Li et al., J Mol Diagn, 2017). The auto-classifier evaluates hotspot frequency (COSMIC, Cancer Hotspots), population frequency (gnomAD, ExAC), functional evidence, and therapeutic associations (CancerKB, CIViC, DrugBank) to assign a preliminary Tier. The scoring criteria and the evidence used for each decision are displayed for pathologist review.
The pathologist can accept the auto-classification or override it, and both actions are logged with a timestamp and user identifier. Auto-classification is a starting point; final sign-off is always a qualified human decision.
Yes. VarSeq supports integrated analysis of the full somatic alteration spectrum within a single sample session: SNVs and indels, somatic CNVs with LOH analysis, gene fusions and structural variants, and direct calculation of TMB (Tumor Mutational Burden) and MSI (Microsatellite Instability) from NGS data.
This matters for labs running comprehensive genomic profiling panels like Illumina TSO-500, where all alteration types contribute to the final clinical report and must be interpreted in the context of one another.
Golden Helix CancerKB is a professionally curated clinical knowledge base developed and maintained by the Golden Helix scientific team. It differs from community-curated databases like CIViC in two ways. First, CancerKB interpretations are tumor-type-specific: a KRAS G12C variant in lung adenocarcinoma carries a different clinical interpretation than the same variant in colorectal cancer, and CancerKB reflects that context explicitly. Second, CancerKB entries include report-ready interpretation language, therapeutic associations with evidence levels, and resistance profiles, pre-synthesized for clinical use rather than requiring the interpreting scientist to synthesize relevance from raw community submissions.
CIViC, COSMIC, and Cancer Hotspots are also integrated in the annotation stack, giving labs both community evidence and curated expert knowledge in the same workflow.
Yes. VSClinical applies AMP classification criteria to somatic variants and ACMG classification criteria to germline variants within the same sample session. This is essential for labs running hereditary cancer panels that detect both tumor-driving somatic mutations and constitutional germline risk variants (BRCA1/2, Lynch syndrome genes, TP53) from a single sequencing assay.
The two classification frameworks are applied separately, with distinct evidence bases and report sections, so somatic and germline findings are not conflated in the final report.
Golden Helix supports on-premises, private cloud, and air-gapped deployment. For oncology labs handling sensitive patient data under HIPAA, CLIA, and institutional data governance requirements, on-premises or private cloud deployment keeps all patient data within the lab's controlled environment. Air-gapped configurations serve facilities with strict network isolation requirements.
All deployment modes support the same VarSeq and VSClinical AMP functionality. See full security and compliance details.
Explore featured articles and expert-led webcasts on precision oncology and somatic interpretation.
AMP-guided somatic interpretation, therapy-matched reporting, and comprehensive tumor profiling in a validated, production-ready platform.