Pipeline integration · interactive

How pipelines run inside SeqDesk

SeqDesk2026-06-30Interactive

SeqDesk runs on a sequencing facility’s own server, so the database, the sequencing files and the analysis all stay in-house. This page walks through what happens when you run a pipeline: how SeqDesk fills in the settings from data it already has, runs the workflow on your own machines, and saves the results back so you can download or share them. Scroll, and each part is shown one step at a time.

The same analysis, two ways

What running one pipeline on a sample takes — by hand, and inside SeqDesk.

By hand

With SeqDesk

Track down each FASTQ file and paste its path into a sheet

The files are already linked to the sample

Look up the platform, organism and grouping, and type them in

Read from the order, sample and study

Log into the server, edit config files, launch Nextflow by hand

Click Run

Dig through output folders and email the files around

Results appear on the sample page, ready to download

Hope the filenames still make sense in six months

Every result stays linked to the sample it came from

Where a pipeline run fits

One sample’s path through SeqDesk, from the lab to the public archive.

Order placed

Samples sequenced

Reads linked

Pipeline runs

Results saved

Submit to ENA

This page is about the highlighted steps — running a pipeline and saving what it produces. Submitting to ENA is its own story.

The reason SeqDesk’s pipelines feel so connected is simple: the data doesn’t move. Everything a pipeline needs is already in the database next to it, and everything it produces goes back there. Start with that database.

The SeqDesk databaseone row per order, sample, read, result

your server · behind your firewall

Study

groups the work

idtitlestudyAccessionId

Order

what came off the machine

idplatform

Sample

the biomaterial

sampleIdtaxIdscientificName

Read

points at the FASTQ files

file1file2checksum1

Assembly

what analysis recovers

assemblyFileassemblyAccessionpreferred

Bin

a genome from the mix

completenesscontaminationbinAccession

pipelinereads these records and asks SeqDesk to write new ones — it never edits the database itself.

It runs on the facility's own server

SeqDesk isn't a service you upload to. A sequencing facility installs it on its own server or cluster. The database and the sequencing files stay there, behind the facility's firewall — nothing is sent outside to be analysed.

Everything is already a row

Each order, sample and read is a record in that database. A sample already carries its details: the organism, the platform it was sequenced on, and the path to its FASTQ files.

Results live here too

When a pipeline produces an assembly or genome bins, those become records as well — linked to the sample they came from.

Pipelines are visitors

A pipeline can read these records and add new ones, but it never edits the database directly. SeqDesk makes every change itself. That keeps your data safe — and it's what makes the next steps work.

Because a sample already lives in that database, picking it as input means SeqDesk can fill in almost everything a pipeline needs to know.

samplesheet.csvmade by SeqDesk

sample*

group*

short_reads_1*

short_reads_2

short_reads_platform*

reference databases — set once by the facility

kraken2Db--kraken2dbhidden from you

gtdbDb--gtdbhidden from you

0from your records0facility configyou type 0

A pipeline needs a settings file

To run, the MAG pipeline needs a samplesheet: which sample, where its read files are, and what platform produced them. Normally you’d fill all of this in by hand.

But SeqDesk already knows

You just pick a sample you entered earlier, and SeqDesk fills every row for you — the file paths come from the read record, the platform from the order, the grouping from the study.

Even the databases are pre-set

Large reference databases — like the one MAG uses to name species — are set once by the facility admin and hidden from your view. You never have to find or type them.

So you type nothing

Every value came from data you already had. You picked a sample; the form filled itself.

With the settings ready, SeqDesk starts the workflow — on your own hardware.

#MAG-20240126-001

pendingqueuedrunningcompleted

Running on the SeqDesk server itself, as a background process.

QC + host removal

Assemble · MEGAHIT

Bin · MetaBAT2 / MaxBin2

Refine · DAS Tool

CheckM · completeness / contamination

Classify · GTDB-Tk

Report · MultiQC

0database writes so far — the workflow only makes files on your disk.

Start the run

SeqDesk builds the command and starts the workflow — either on the SeqDesk server itself, or on your computing cluster. Use the switch to see both.

Follow along

The run moves through clear stages, and you can watch each step — assembly, binning, naming the species — light up as it happens, instead of reading raw logs.

Nothing has been saved yet

While it runs, the pipeline only writes files on your disk. The database hasn't changed at all — not one record. Saving the results is a separate, careful step.

When the workflow finishes, it has only made files. Turning those files into records the rest of SeqDesk can use is the last step — and SeqDesk does it carefully.

Results saved as recordsSeqDesk does the writing

FASTQ ChecksumRead.checksum1saved

FastQCRead.fastqcReport1saved

MAGAssembly.assemblyFile · Bin.completenesssaved

SubMGAssembly.assemblyAccession · Bin.binAccessionsaved

Read Cleaningstaged for reviewneeds approval

Cleaned-up reads wait for a check. They don’t overwrite your originals — an admin approves them first.

Now it’s yours to use

Download assemblyView HTML reportPublish to ENAShare with the study

SeqDesk saves the simple results

When the run finishes, SeqDesk finds the output files, matches them to your samples, and saves the small results — checksums and quality numbers — straight onto the read records.

And the big ones

Assemblies and genome bins become their own records, linked to the sample. If you submitted to ENA, the accession numbers are saved back onto them too.

Some results wait for a check

Cleaned-up read files are different: they don't overwrite your originals. They wait for an admin to approve them, so nothing is lost by accident.

Now it's yours to use

The results show up on the sample and study pages. You can download an assembly, open a report, submit to ENA, or share with your team.

That’s the whole loop. The data never leaves the facility, the sample’s own records fill in the run, and the results come back to the same place you started — ready to download, submit, or share. The best way to see it is to try it.

Live Demo

Try SeqDesk with sample data

Explore a fully functional workspace pre-loaded with sample sequencing orders, studies, and submissions. Switch between the researcher and facility admin views.

Live demo is best on desktop

The interactive demo requires a larger screen. Open this page on a desktop browser or try the full demo link below.

Open full demo

Preparing a shared demo workspace…

Both live app views attach to the same disposable demo data.

Sources. Built from SeqDesk’s pipeline documentation — pipeline scopes, running pipelines, adding pipelines, available pipelines and results. The pipeline names, record fields, samplesheet columns, run stages and writeback rules come from those pages; the per-pipeline result-type labels are the docs’ exact pair for Read Cleaning and modelled from documented behaviour for the rest. The sample id, file paths and run number shown are examples, not real records. For how SeqDesk submits to ENA, see Up the wire.