NGS Cost per Sample in 2026: Drivers Beyond Instrument Fees

NGS cost per sample is the all-in figure that covers one library from preparation through sequencing, analysis, and data retention. Instrument or flow-cell fees are only one line. How much next generation sequencing costs in 2026 still depends on prep, multiplex, depth, re-runs, bioinformatics, and storage.

Core labs and commercial providers split those extras differently. Compare quotes by included steps, not by the sequencer model alone.

Why Instrument Fees Are Only One Line of NGS Cost per Sample

Sequencers are shared machines. The run fee pays for a flow cell, a reagent kit, and instrument time. That fee is then divided among the libraries loaded on the run. A sample's share moves when pooling density changes, when a dedicated flow cell is requested, or when unused occupancy cannot be filled. The instrument sticker therefore never equals the sample invoice unless the quote already rolled prep, analysis, and storage into one bundled number.

How much next-generation sequencing costs per sample is a workflow question. Extraction, fragment or amplicon prep, adapter ligation, indexing, quantification, sequencing, demultiplexing, optional alignment or counting, delivery, and retention are separate cost objects. Two labs can sequence on the same instrument class and still see unlike invoices because they stopped the workflow at different steps. The evaluation axis is the finished data product the project needs. The practical direction is to name that product (raw FASTQ, aligned reads, counts, or a variant table) before anyone talks about a sequencer.

Sanger work is billed on a different unit (one primer-template injection) and is out of scope here. NGS quotes should be compared to other NGS quotes that deliver the same data product, not to a capillary reaction rate.

Library Preparation as the First Multiplier

For small batches, library prep often rivals or exceeds the sequencing-reagent line. Prep is labor, kit chemistry, dual indexing, size selection, and quality control. RNA-seq adds depletion or poly(A) selection. Target enrichment adds probes and extra PCR. Amplicon panels look cheap until primer pools, unique dual indexes, and failed-library repeats are counted. A quote that lists only "sequencing" is usually incomplete if the core or vendor is also building the library.

Indexing is easy to miss. Unique dual indexes, unique molecular identifiers, and reserved barcode collisions each change kit choice and the number of samples that can share a run. Customer-supplied indexes can reduce a line item or create a failed-demux charge if the plate map is wrong. Quantification (fluorometric assay, electrophoresis, or qPCR of the finished library) is another add-on that decides whether a library is loaded at all.

• Extraction and QC of input nucleic acid: poor input creates failed libraries that are billed as prep even when no useful reads are produced.
• Library kit and protocol class: DNA-seq, RNA-seq, amplicon, and hybrid-capture kits have different reagent and labor loads.
• Index strategy: unique dual indexes and UMIs add kit cost but reduce demultiplex risk on shared flow cells.
• Library quantification and size checks: loading decisions depend on these assays, which may appear as separate QC fees.
• Failed-library repeats: a second prep on the same sample is often a new billable library, not a free continuation of the first.

The evaluation axis is successful libraries ready to load, not samples received. The practical direction is to ask whether prep, indexes, and QC are included, optional, or customer-supplied, and what happens when a library fails quality gates before the run.

Depth Targets, Multiplex Batching, and Shared Runs

Requested depth is a billing lever, not a biology tutorial. More requested reads means a larger share of the flow cell, more cycles, or a second run. Teams set a coverage or read-count target because the scientific question needs a certain amount of data. The invoice follows that request. Over-requesting depth raises cost per sample without changing the instrument model. Under-requesting depth creates a later top-up run that is usually billed as new sequencing, not as a correction of the first quote.

Multiplex is the usual way to lower NGS cost per sample: several indexed libraries share one run. The discount is real only when the pool is balanced, indexes are unique, and the lab can wait for a full batch. A half-filled run still consumes reagents. A rushed dedicated flow cell raises the per-sample share. Pooling imbalance can starve one sample of reads while another is over-sequenced, which may trigger a paid re-run of the weak library.

Do not treat "coverage" and "depth" as interchangeable invoice words. Ask the vendor whether they are selling a read-count guarantee, a lane share, or a best-effort placement in the next multiplex. The evaluation axis is data delivered to the stated target, including any top-up policy. The practical direction is to price the read budget you will actually analyze, then ask how multiplexing and wait time change that number.

Bioinformatics, Re-runs, and Failed Libraries

Demultiplexing to FASTQ is often included. Alignment, duplicate marking, variant calling, RNA-seq counting, and custom reports often are not. Cloud compute, reference-genome licenses, and analyst time sit behind those extras. A lab that only needs FASTQ should not pay for a pipeline it will rerun internally. A lab that needs a locked report should not assume FASTQ delivery includes interpretation. Spell the handoff format and whether BAM, CRAM, or count tables are retained after delivery.

Re-runs split into three causes, and they are billed differently. A chemistry or instrument failure is commonly repeated without a new sequencing charge. A library that was under-quantified, contaminated, or low-complexity is commonly billed again, including a second prep. A scientific request to add depth after seeing the first results is a new order. Mixing these three in one "repeat" budget line makes the project miss the true NGS cost per sample.

Keep sample identifiers, index IDs, library lots, and run IDs in the experiment record so a billed sample maps to the aliquot that was shipped. An electronic lab notebook is the right place for that chain when several people handle extraction, pooling, and data return. The evaluation axis is cost per accepted dataset, including analysis and repeats. The practical direction is to demand written rules for each failure class before the first sample is extracted.

Data Storage, Delivery, and Retention

NGS invoices increasingly separate sequence generation from data custody. FASTQ and BAM files are large. Delivery may use a portal, a cloud bucket, or physical media, and egress can be a billed line. Retention is time-limited in many cores: after the stated window, the lab either downloads, pays for extended storage, or loses the primary files. Secondary analysis copies can double the stored volume if BAM and FASTQ are both kept.

Storage is not a sequencing science problem, but it is part of NGS cost per sample for any project that must remain reanalyzable. Ask who holds the primary files, for how long, in what format, and what restoration costs after deletion. If the lab will move files into its own archive, budget transfer time and a checksum step. Sequence design files and sample metadata should travel with the run IDs so a later reanalysis still knows which construct or amplicon the library was meant to represent. Standard molecular biology tools can hold those construct and primer records next to the sample names used on the pooling sheet.

The evaluation axis is the cost to keep a usable dataset for the project's required life, not the cost to generate reads for one week. The practical direction is to put retention, format, and egress on the quote beside prep and sequencing.

What an NGS Quote Should Itemize in 2026

Use the table as a line-item checklist when a vendor or core says they have priced "per sample." It is not a rate card and it does not rank providers.

Map every row onto the assay you are actually running. Whole-genome, whole-exome, RNA-seq, and small amplicon panels do not share a prep or a read budget, so they should not share an unstated "typical sample" price. If a row is blank, treat it as a later invoice, not as included work.

FAQ

What is included in NGS cost per sample?

Only the steps the quote names. A complete per-sample figure usually has to cover input QC, library prep, indexes, quantification, a defined read budget, demultiplexing, delivery, and a retention window. Many quotes include only sequencing reagents and instrument time, then add prep, analysis, and storage later. Ask for the data product being sold: raw FASTQ, aligned files, or an interpreted report. Ask how a failed library or a missed read target is billed. Two providers can sequence similar libraries and still invoice different totals because they stopped at different steps. Write those steps down before you compare unit prices. The useful number is cost per accepted dataset, not cost per tube that entered the lab.

Does next-generation sequencing pricing include library preparation?

Not unless the quote says so. Library prep is a separate kit-and-labor process: fragmentation or amplicon generation, adapter ligation, indexing, amplification, and QC. RNA-seq and hybrid-capture protocols add further reagent lines. Some cores price prep and sequencing as a bundle; others allow customer-made libraries and charge only for loading and the run. Customer-made libraries shift cost to the lab but do not remove failed-library risk, because a poor library can still occupy a paid share of the flow cell. Confirm who supplies unique dual indexes and who pays if the library fails quantification. If prep is included, ask whether a failed prep is repeated inside the bundle or billed as a second library.

How does multiplexing change NGS cost per sample?

Multiplexing spreads one run's reagent and instrument cost across indexed libraries, which usually lowers the per-sample sequencing line. The saving holds when indexes are unique, the pool is balanced, and the project can wait for a full batch. It shrinks when the lab requests a dedicated flow cell, ships too few samples to fill a pool, or accepts imbalance that leaves one sample under-sequenced. Under-sequenced samples often return as paid top-ups. Multiplex also adds index-kit cost and demultiplex risk, so the lowest sequencing line is not automatically the lowest sample total. Ask for the minimum pool size, the wait policy, and the fee for a re-pool. Price the read budget each sample needs after pooling, not only the shared run fee.

Is bioinformatics included in an NGS sample quote?

Demultiplexing to FASTQ is commonly included. Alignment, variant calling, RNA-seq quantification, annotation, and custom reports commonly are not. Those steps add compute, reference resources, and analyst time, and they change the legal handoff of what the vendor is claiming. A lab with its own pipeline should insist on FASTQ plus a clear index manifest and should not pay for unused interpretation. A lab that needs a locked analysis should specify the pipeline version, reference, and deliverable files. Storage of BAM alongside FASTQ can add a retention fee. Ask whether reanalysis after a pipeline update is included. Treat bioinformatics as its own line so sequencing chemistry and interpretation are not confused on the invoice.

Who pays for NGS re-runs and failed libraries?

Instrument or reagent failure is commonly re-run without a new sequencing charge. Libraries that fail QC because of input quality, low complexity, contamination, or index collision are commonly billed again, including a second prep if one is needed. Adding depth after seeing the first results is a new order, not a warranty claim. The quote should separate those three classes in writing. It should also say whether unused reads from an over-sequenced sibling in the same pool can be credited. Keep library lot, index ID, and concentration with the sample record so a dispute is factual. A silent re-run policy makes NGS cost per sample impossible to forecast, because one failed plate can exceed the original sequencing line.

Do NGS vendors charge for data storage after the run?

Many do, or they include only a short retention window and then delete primary files. FASTQ and BAM volumes are large enough that portal hosting, cloud buckets, and egress can appear as explicit fees. Some cores treat download as included during the window and charge only for extension. Others bill storage by gigabyte-month or by project. Ask how long files remain, which formats are kept, whether a checksum is provided, and what restoration costs after deletion. If the project must remain reanalyzable, budget a lab-side archive and a documented run ID. Storage is part of the sample cost whenever the scientific record depends on the primary reads, not only on a summary table.

Conclusion

NGS cost per sample in 2026 is driven by library prep, indexes, requested depth, multiplex occupancy, bioinformatics, re-run rules, and data retention, not by instrument fees alone. Compare quotes as itemized workflows that end in a named data product. For teams that need sample, index, and construct identifiers recorded beside the run, keep those records in one research workspace such as Zettalab so the invoice, the library, and the files still refer to the same sample.