Genomics

Lineages,assigned offline

Pango lineage assignment over consensus genomes.The lineage data is in the image, so nothing is fetched.

Pangolin 4.3.44 cores · 8 GBRuns as a job

hub://pangolin

$ dxflow workflow create --identity pangolin hub://pangolin --start

01Pick a machinecores, memory and a rate per hour

02Create the workflowpulled from the hub, once

03Run it, then stopstopped when the work is done

The application

A FASTA in,a report out

One row per sequence, with the lineage, the conflict and the QC.

Consensus genomes

One FASTA of near-complete SARS-CoV-2 genomes is what it expects.

One CSV back

lineage_report.csv, a row per sequence, and the versions it used.

Nothing to download

The pdata is baked in, so assignment runs with no network at all.

How you run it

One FASTA in,one report out

Consensus genomes go up, a lineage call per sequence comes back.

INPUTThe query FASTA. One genome or a few hundred, in a single file.

THREADSHow much of the machine the assignment gets to use.

EXTRAFurther pangolin flags, such as choosing the usher or fast analysis mode.

a run, end to end

$ dxflow artifact upload sequences.fasta input/

$ dxflow workflow start pangolin --override env.job.EXTRA=--analysis-mode=usher

$ dxflow artifact download output/ ./lineages/

The job

It runs,and then it stops

A job, not a session — it ends when the report is written.

inputInputyour genomes, read-only

outputOutputthe lineage report

Point INPUT at itThe default is /data/input/sequences.fasta — override it.

Watch it from the CLIdxflow workflow logs follows the run, and list shows the state.

Good to know

It carries its owndesignation data

The lineage data is baked into the image, so the call happens without reaching for the network.

Offline by build

pdata ships inside the image. No download at start, and the same image gives the same call tomorrow.

Read qc_status first

Short or thin coverage still gets a row, marked fail. Near-complete genomes are what it is built for.

The CSV it writes

Lineage, conflict and QC per sequence, in one table you can join against your metadata.

The image

Pulled once,then it stays

Pangolin arrives as one image. This is what comes down the first time, and what the disk should have free for it.

891Mamd64compressed, the way the registry counts it

10GOn diskunpacked, with room to work beside it

ghcr.io/dxflow-ai/pangolin:latestPublished from the hub, pulled on the first start and kept for the ones after it.

amd64 onlyOne architecture is published, so pick a machine that matches it.

What it asks for

What it wants,and what it needs

The definition asks for 4 cores and 8 GB. The image comes up on less than that, and a start given --fit trims the ask to whatever the machine actually has.

4 cores · 8 GBAsks forwhat the definition writes down

2 cores · 4 GBRuns onthe least the image comes up on

Not neededGPUit works on the cores alone

The ask is not the floorThe definition writes down what suits the work. The image itself starts on less, which is what the second figure is.

--fit caps it to the hostA start given --fit trims each step to what the machine actually has, for that start alone. The definition is never rewritten.

Machines that fit it

Pangolin asks for 4 cores and 8 GB. Cheapest first.

E2 Standard-4

$0.168/ hour4 cores · 16 GBStart this machine

B4ms

$0.208/ hour4 cores · 16 GBStart this machine

T3 XLarge

$0.208/ hour4 cores · 16 GBStart this machine

Run Pangolin on your own machinePick a machine that meets it, and it opens about a minute after you ask.

Start a machine Browse workflows