NGS Workflow End-to-End: From Sample to Variant Report

Next-generation sequencing turns a tube of DNA into millions of short reads, and ultimately into a clinical variant report. Between those two points sits a carefully controlled pipeline: accessioning, nucleic acid extraction, library preparation, sequencing, secondary bioinformatics, tertiary interpretation, and reporting — each with its own quality control checkpoints. Understanding the NGS workflow end to end helps lab teams see where errors creep in, where automation and LIMS software add value, and where qualified human review must always remain in command. This guide walks the full pipeline from sample to signed-out report.

Overview: what “end to end” really means

An NGS workflow is best understood as two halves joined in the middle. The wet lab turns a physical specimen into sequence data; the dry lab turns that data into an interpreted, reportable result. Quality control runs throughout both, and a LIMS or hybrid LIS/LIMS ties the stages together so every sample, reagent lot, instrument run, and QC metric is traceable back to the accession. Below, we follow a single sample through the pipeline.

1. Accessioning and sample receipt

The journey starts when the specimen arrives. Accessioning assigns a unique identifier, records sample type and condition, verifies the order and consent, and logs chain of custody. This is the anchor point for traceability: every downstream plate, library, and read should link back to this accession.

QC at this stage: specimen suitability, labeling integrity, and rejection criteria (e.g., insufficient volume, improper transport). Catching problems here prevents wasted reagents and ambiguous results later.

2. Nucleic acid extraction

DNA or RNA is isolated from the specimen. Yield and purity determine whether the sample can proceed.

QC at this stage: concentration (fluorometric quantification), purity ratios, and — for RNA or degraded samples — integrity metrics. Samples that fail thresholds are re-extracted or flagged.

3. Library preparation

Extracted nucleic acid is converted into a sequencing library: fragmentation, adapter ligation, indexing (barcoding), and often target enrichment for panels or exomes. Indexing is what lets many samples share a single sequencing run, so accurate index assignment and plate mapping are critical.

QC at this stage: library concentration, fragment size distribution, and — where used — enrichment performance. A LIMS recording plate maps, well positions, and reagent lots here prevents sample swaps and supports lot-level traceability.

4. Sequencing

The pooled libraries are loaded onto a sequencer, which generates raw reads. Run configuration (read length, depth targets) is set according to the validated assay.

QC at this stage: run-level metrics such as cluster density, base-call quality scores, yield, and error rates. These confirm the run itself performed within validated specifications before any sample-level analysis is trusted.

5. Secondary analysis (bioinformatics)

Raw reads enter the bioinformatics pipeline. Secondary analysis typically includes quality trimming, alignment to a reference genome, and variant calling — producing SNVs, indels, and, depending on the assay, copy-number or structural variants.

QC at this stage: per-sample coverage and depth, uniformity, mapping rates, and on-target percentages. Regions failing coverage thresholds are flagged as not adequately assessed — an important limitation that must carry through to the report.

6. Tertiary analysis and variant interpretation

This is where data becomes meaning. Called variants are annotated (population frequencies, in-silico predictions, literature and database evidence), filtered against the assay's scope, and classified for clinical significance. For pharmacogenomics, genotypes are translated into predicted metabolizer phenotypes using established guidelines and the lab's validated logic.

Here the human role is paramount. Bioinformatics and AI-assisted tools can rank candidates, pre-fill annotations, surface relevant evidence, and draft narrative language to speed the work. They do not assign final clinical classification or sign out a case. Qualified laboratory scientists and directors review the evidence, apply validated criteria, resolve ambiguity, and take responsibility for the interpretation. Software accelerates and documents this review; it never replaces it.

QC at this stage: confirmation of low-confidence calls (e.g., orthogonal or Sanger confirmation per the lab's validated policy), review of borderline classifications, and verification that filtering didn't exclude reportable findings.

7. Reporting

The approved interpretation is assembled into a structured clinical report: result summary, variant or genotype tables, predicted phenotypes for PGx, methodology, limitations (including regions of inadequate coverage), and clinical context. The report should support versioning and amendments, with a clear audit trail of who reviewed and approved it — the focus of Labrynix Reports.

QC at this stage: final review and sign-out by an authorized individual, and verification that the report accurately reflects the validated result and its limitations.

8. Delivery and downstream

Finalized reports are delivered to ordering providers and, where appropriate, patients — and, in parallel, the case feeds billing and revenue cycle workflows. Closing this loop keeps operations, clinical results, and reimbursement aligned.

Where software and LIMS fit

Across all eight stages, a LIMS or hybrid LIS/LIMS provides the connective tissue:

• Sample, plate, and run tracking so every read traces back to a tube and an accession.
• Reagent and lot management for traceability and recalls.
• Instrument integration to capture run metrics and reduce transcription error — see Labrynix Connect.
• QC capture and gating so failed metrics are visible and acted on, not buried in instrument exports.
• Interpretation and reporting support that organizes evidence and drafts documentation for qualified review.
• Audit trails that record who did what, when — essential for accreditation and troubleshooting.

The goal isn't to remove people from the pipeline. It's to remove transcription errors, lost samples, and untracked QC — so scientists spend their time on judgment, not janitorial data work. That is the role of the Labrynix LIMS within the wider platform.

A note on QC discipline

QC is not a single gate at the end; it's a chain. A sample that passes extraction can still fail at library prep; a run that passes run-level metrics can still have an individual sample with inadequate coverage. Treating each checkpoint as independent — and recording the results — prevents a single masked failure from reaching a report. The strongest NGS operations make every QC decision explicit, traceable, and reviewable. For labs built around this work, see our molecular diagnostic lab solutions.