Next Generation Sequencing Service: A Procurement Guide (Including 16S rRNA Sequencing)

Outsourcing sequencing to a next generation sequencing (NGS) service provider — rather than running samples on an owned instrument — is now the default path for most labs that need sequencing occasionally, at a scale their own core can’t support, or on a platform they don’t operate. This guide is written for the lab manager, principal investigator, or procurement officer who has to select and vet an NGS service vendor: what to specify in a request for quote, which deliverables and quality metrics to require in writing, what accreditation actually means for a given use case, and how to evaluate a 16S rRNA amplicon sequencing service specifically, since that is one of the most commonly outsourced NGS applications and has its own set of procurement questions.

What “NGS service” Actually Covers

“Next generation sequencing service” is a broad commercial category, not a single product. Before writing a quote request, narrow down which of these you’re actually buying, because pricing, turnaround, and the right accreditation questions differ across them:

• Fee-for-service sequencing only. You supply an extracted, quantified, quality-checked nucleic acid sample; the vendor runs library preparation and sequencing and returns raw or demultiplexed reads (FASTQ). This is the narrowest, cheapest, fastest-turnaround service tier.
• Library prep + sequencing. The vendor takes your DNA/RNA sample from extraction or purification through sequencing-ready library construction. Adds cost and a quality-control checkpoint you don’t control directly (library QC before sequencing), but removes a common failure point for labs without in-house library-prep expertise.
• End-to-end service including bioinformatics. Sequencing plus a defined analysis pipeline (e.g., variant calling, differential expression, taxonomic classification for amplicon data). This is where deliverable specification matters most — “bioinformatics included” can mean anything from a standard automated report to a fully customized analysis, and vendors differ enormously in what’s actually included at the quoted price versus billed as an add-on.
• Full project design and consulting. The vendor helps design the experiment (platform choice, read depth, replicate number) as well as executing it. Useful for labs sequencing outside their core expertise, but blurs who is accountable for a study being adequately powered.

Get the tier explicit in writing before comparing quotes across vendors — a lower price on a “sequencing only” quote is not directly comparable to a higher price that includes library prep and analysis.

Core Evaluation Criteria for an NGS Service Vendor

Platform and chemistry

Confirm which sequencing platform(s) the vendor actually operates in-house versus subcontracts. Read length, error profile, and throughput differ meaningfully between short-read platforms (e.g., Illumina) and long-read platforms (e.g., PacBio, Oxford Nanopore), and the right choice depends on your application — short, high-accuracy reads for variant calling or amplicon sequencing; long reads for structural variant detection, de novo assembly, or full-length transcript/operon sequencing. A vendor quoting a service without specifying the platform and read configuration (read length, single- or paired-end) hasn’t given you enough information to evaluate the quote against a competitor’s.

Accreditation — match it to what the data is actually used for

This is the most commonly misunderstood procurement question in NGS services, so it’s worth stating precisely:

• CLIA (Clinical Laboratory Improvement Amendments) certification is a US federal requirement that applies to laboratories testing human specimens for the diagnosis, prevention, or treatment of disease, or for the assessment of health — i.e., results that go back to a patient or their treating clinician. A sequencing run generating research-only data on de-identified or non-clinical samples does not require a CLIA-certified lab. If you are procuring sequencing whose results will inform a clinical decision, or that will be reported into a patient’s medical record, the vendor must operate under CLIA certification (or an equivalent state licensure program in states like New York that run their own oversight in place of, or alongside, CLIA).
• CAP (College of American Pathologists) laboratory accreditation is a voluntary, more detailed accreditation program that many clinical labs use to demonstrate compliance with CLIA (CAP-accredited labs are deemed to meet CLIA requirements) and that adds requirements beyond the CLIA floor. For clinical-grade sequencing — e.g., a hereditary cancer panel, an oncology companion-diagnostic assay — ask specifically whether the vendor is CAP-accredited for the specific test, not just “CLIA-certified,” since CLIA certification alone is a lower bar.
• ISO/IEC 17025 accreditation covers testing and calibration laboratory competence generally and is a relevant credential to ask about for a non-clinical NGS service vendor (e.g., environmental or agricultural sequencing) where CLIA doesn’t apply but you still want independent evidence of a validated quality system.
• Research-use-only (RUO) sequencing — the large majority of academic core-facility and commercial sequencing-service work — does not require any of the above by regulation. Vendors should still be able to describe their internal quality-management system (run-QC metrics tracked, instrument maintenance/calibration schedule, sample-tracking chain of custody) even without formal accreditation; ask for it in writing as part of vendor qualification, not as an afterthought.

The procurement error to avoid in both directions: over-specifying CLIA/CAP for research-only work (which needlessly restricts your vendor pool and raises price with no data-quality benefit), and under-specifying it for anything whose output will touch a clinical or regulatory decision.

Sample requirements and quality control

Every vendor publishes minimum sample requirements — concentration, volume, purity ratios (A260/280, A260/230), and integrity metrics (RIN for RNA, DIN for DNA). Get these in writing before shipping samples; a sample that fails a vendor’s incoming QC either gets rejected (delay, and sometimes a re-extraction cost on your side) or run anyway with a documented QC exception that should be disclosed in the final report. Ask explicitly what happens in the exception case — some vendors proceed and flag it, others require sign-off before proceeding, and a few will run it without telling you, which is a real quality risk worth screening out in the RFQ stage.

Turnaround time and batching

Quoted turnaround time is typically measured from receipt of QC-passing sample to data delivery, and commonly excludes the sample-shipping and incoming-QC period — read the fine print. Turnaround is also frequently tied to batch scheduling: many providers run set flow-cell configurations on a weekly or biweekly cadence, so a single sample submitted the day after a batch closes can add a full cycle to your actual timeline versus the vendor’s headline turnaround figure. If timeline is a real constraint (e.g., a grant deadline or a clinical trial protocol window), ask for the vendor’s batching schedule, not just the advertised turnaround number.

Data delivery, retention, and ownership

Confirm in the service agreement: file format(s) delivered (raw BCL vs. demultiplexed FASTQ vs. a processed output), delivery mechanism and its security posture (a secure transfer portal vs. physical drive), how long the vendor retains raw data and libraries after delivery (relevant if you need a re-run or a deeper analysis later without re-sequencing), and — for any project involving human subject data — whether the vendor’s data-handling terms are compatible with your institution’s IRB protocol and any applicable data use agreement.

Cost structure

Per-sample sequencing cost is driven primarily by sequencing depth (reads per sample) and, for platforms like Illumina, how many samples are multiplexed onto a shared flow cell — a full flow cell split across more samples lowers per-sample cost but also lowers per-sample depth. Get quotes normalized to a specific depth target (e.g., cost per sample at 20 million read pairs) rather than comparing headline per-sample prices that may assume different depths. As with the DNA sequencer capital-purchase side of this decision, most NGS service vendors do not publish a fixed price list online and expect a quote request specifying sample number, application, and depth target — build your RFQ around those three numbers to get comparable quotes across vendors.

16S rRNA Sequencing Service: What to Specify

16S rRNA gene sequencing — targeted amplicon sequencing of the bacterial and archaeal ribosomal RNA gene used for microbial community profiling — is one of the most frequently outsourced NGS applications, and it has its own procurement checklist beyond the general criteria above.

• Hypervariable region coverage. The 16S rRNA gene contains nine hypervariable regions (V1-V9) interspersed with conserved regions used for primer binding. Most short-read amplicon services sequence one or two adjacent regions rather than the full gene — the V3-V4 region is the most commonly offered target on Illumina paired-end platforms because it balances taxonomic resolution against the read-length limits of that chemistry. Confirm exactly which region(s) and which primer set the vendor uses, since this affects which taxa are resolved well and directly determines whether your results are comparable to other studies or prior data in your own project that used a different region.
• Full-length 16S option. Long-read platforms (PacBio, Oxford Nanopore) can sequence the full ~1,500 bp 16S gene in one read, generally improving species-level taxonomic resolution over short-read, single-region approaches. If species-level (rather than genus-level) resolution matters for your study, ask specifically whether the vendor offers a full-length option and what platform it runs on.
• Reference database and bioinformatics pipeline. If the service includes taxonomic classification, ask which reference database it uses (commonly SILVA, Greengenes, or the RDP database) and which version — database choice and version measurably affect the taxonomic calls returned, and reproducing or comparing results across studies requires knowing which one was used.
• Negative and mock-community controls. Because 16S amplicon sequencing is highly sensitive to low-biomass contamination (reagents, extraction kits, and lab surfaces all carry trace bacterial DNA), ask whether the vendor runs extraction/no-template negative controls and, ideally, a mock community of known composition as a positive control, and whether those controls are included in your per-sample price or must be ordered separately.
• Metagenomic shotgun sequencing as an alternative. If genus-level community profiling isn’t sufficient — for example, if you need functional gene content or strain-level resolution — ask the vendor whether they also offer shotgun metagenomic sequencing, which sequences all DNA in a sample rather than a single targeted amplicon, at correspondingly higher cost and data volume per sample.

Building a Comparable RFQ

To get quotes from multiple vendors that can actually be compared line-for-line, specify in every request:

1. Application (e.g., whole-genome sequencing, targeted panel, 16S amplicon V3-V4, RNA-seq) and organism/sample type.
2. Sample number, and current sample QC metrics if already known (concentration, A260/280, RIN/DIN).
3. Requested platform, or “vendor’s recommendation” if you’re open to it — but then require the vendor to state which platform they’d use.
4. Target sequencing depth per sample.
5. Deliverable tier (sequencing only / library prep + sequencing / + bioinformatics) per the categories above.
6. Required turnaround, stated as calendar days from QC-passed sample receipt to data delivery, not from order placement.
7. Accreditation requirement, if any (CLIA/CAP for clinical-use data; none required for RUO research use, but request a description of the internal QC system regardless).
8. Data delivery format and retention period.

Institutions with an established procurement office may also want to check whether an existing standing contract or preferred-vendor agreement already covers sequencing services — see CASRAI’s broader guidance on evaluating a DNA sequencer purchase for the related capital-equipment decision (buying an instrument outright) that a high sequencing volume may eventually justify instead of continued outsourcing.

Frequently Asked Questions

Does an NGS service provider need to be CLIA-certified?

Only if the results will be used for clinical diagnosis, treatment decisions, or reporting into a patient’s medical record. Research-use-only sequencing on de-identified or non-clinical samples does not require CLIA certification. If your project has any clinical-reporting component, confirm CLIA certification (and, for higher-stakes clinical tests, CAP accreditation for that specific assay) before sending samples.

What’s the difference between a 16S rRNA sequencing service and shotgun metagenomic sequencing?

16S rRNA sequencing amplifies and sequences one or two hypervariable regions of a single conserved bacterial/archaeal gene to profile which taxa are present, typically to the genus level. Shotgun metagenomic sequencing sequences all DNA in a sample, which costs more and produces more data per sample but can resolve species/strain-level identity and recover functional gene content, not just taxonomic composition.

Which 16S region should I choose if I have no prior constraint?

V3-V4 is the most widely used region on short-read platforms because of its balance of taxonomic informativeness and compatibility with common paired-end read lengths, and using it makes your data easier to compare against the largest body of existing published community-profiling data. If a specific prior dataset in your project used a different region, match it rather than defaulting to V3-V4, since results across different 16S regions are not directly comparable.

How is NGS service pricing normally structured?

Per-sample cost, scaled primarily by requested sequencing depth and how many samples are multiplexed together on a shared sequencing run. Vendors rarely publish a fixed price list for anything beyond a small standardized package; expect to submit an RFQ specifying sample number, application, and depth to get an accurate quote.

Do I need to specify a bioinformatics pipeline, or can the vendor choose one?

Specify it, or at minimum require the vendor to document exactly which pipeline, reference database, and version they use as a deliverable. This matters both for reproducibility and because a change in pipeline or database version between two rounds of sequencing for the same project can shift taxonomic or variant calls independent of any real biological difference in the samples.