COVID-19 PubSeq: Public SARS-CoV-2 Sequence Resource

public sequences ready for download!

Edit text!


1 What is the 'public sequence resource' about?

PubSeq, the public sequence resource, aims to provide a generic and useful resource for COVID-19 research. The focus is on providing the best possible sequence data with associated metadata that can be used for sequence comparison and protein prediction.

Just to avoid the notion that PubSeq is a biorepository: it is different. We are not competing with the likes of GenBank and EBI/ENA Open Data - in fact, PubSeq federates open data and adds useful normalised metadata and opens up running workflows by anyone on that data.

Note that there is no conflict of posting data to multiple repositories.

It is better to view PubSeq as an open precision medicine initiative that allows for early detection of new variants and strains; very valuable for both testing and treatment. We have live analysis work flows that run on uploading a new sequence. Contributing sequences helps the overall detection. At this moment the output is limited. We will soon add interactive discovery tools. Uploading data from a sequencer will give researchers all information they want in 5 hours without any bioinformatics knowledge. We may even provide workflows that take data straight from the sequencer. In fact, you get very little out of existing biorepositories, in our opinion, unless you have solid bioinformatics support. And even then there is the question about comparing data that was created using different technologies and workflows. How do you deal with that?

With PubSeq we are in a position to reanalyse that data from raw material when new insights arise. It is likely that COVID19 will be around for another year at least. After that we should prepare for the next pandemic. PubSeq is meant for that. We take the long view and it will only get more powerful. Who knows: maybe the live analysis part of PubSeq will be an approach that other biorepositories will follow.

2 Presentations

We presented at the BOSC 2020 Have a look at the video (alternative link) and the poster.

3 Who created the public sequence resource?

The public sequence resource is an initiative by bioinformatics and ontology experts who want to create something agile and useful for the wider research community. The initiative started at the COVID-19 biohackathon in April 2020 and is ongoing. The main project drivers are Pjotr Prins (UTHSC), Peter Amstutz (Curii), Andrea Guarracino (University of Rome Tor Vergata), Michael Crusoe (Common Workflow Language), Thomas Liener (consultant, formerly EBI), Erik Garrison (UCSC) and Jerven Bolleman (Swiss Institute of Bioinformatics).

Notably, as this is a free software initiative, the project represents major work by hundreds of software developers and ontology and data wrangling experts. Thank you everyone!

4 How does the public sequence resource compare to other data resources?

The short version is that we use state-of-the-art practices in bioinformatics using agile methods. Unlike the resources from large institutes we can improve things on a dime and anyone can contribute to building out this resource! Sequences from GenBank, EBI/ENA and others are regularly added to PubSeq. We encourage people to everyone to submit on PubSeq because of its superior live tooling and metadata support (see the next question).

Importantly: all data is published under either the Creative Commons 4.0 attribution license or the CC0 “No Rights Reserved” license which means it data can be published and workflows can run in public environments allowing for improved access for research and reproducible results. This contrasts with some other public resources, as described in this Nature article.

5 Why should I upload my data here?

  1. We champion truly shareable data without licensing restrictions - with proper attribution
  2. We provide full metadata support using state-of-the-art ontology's
  3. We provide a web-based sequence uploader and a command-line version for bulk uploads
  4. We provide a live SPARQL end-point for all metadata
  5. We provide free data analysis and sequence comparison triggered on data upload
  6. We do real work for you, with this link you can see the last run took 5.5 hours!
  7. We provide free downloads of all computed output
  8. There is no need to set up pipelines and/or compute clusters
  9. All workflows get triggered on uploading a new sequence
  10. When someone (you?) improves the software/workflows and everyone benefits
  11. Your data gets automatically integrated with the Swiss Institure of Bioinformatics COVID-19 knowledge base (Elixir Switzerland)
  12. Your data will be used to develop drug targets

Finally, if you upload your data here we have workflows that output formatted data suitable for uploading to EBI resources (and soon others). Uploading your data here get your data ready for upload to multiple resources.

6 Why should I not upload by data here?

Funny question. There are only good reasons to upload your data here and make it available to the widest audience possible.

In fact, you can upload your data here as well as to other resources. It is your data after all. No one can prevent you from uploading your data to multiple resources.

We recommend uploading to EBI and NCBI resources using our data conversion tools. It means you only enter data once and make the process smooth. You can also use our command line data uploader for bulk uploads!

7 How does the public sequence resource work?

On uploading a sequence with metadata it will automatically be processed and incorporated into the public pangenome with metadata using workflows from the High Performance Open Biology Lab defined here.

8 Who uses the public sequence resource?

The Swiss Institute of Bioinformatics has included this data in the SPARQL endpoint.

The Pantograph viewer uses PubSeq data for their visualisations.

UTHSC (USA), ESR (New Zealand) and ORNL (USA) use COVID-19 PubSeq data for monitoring, protein prediction and drug development.

Amazon AWS has included PubSeq in the Amazon AWS OpenData registry.

9 How can I contribute?

You can contribute by submitting sequences, updating metadata, submit issues on our issue tracker, and more importantly add functionality. See 'How do I change the source code' below. Read through our online documentation at as a starting point.

10 Is this about open data?

All data is published under a Creative Commons 4.0 attribution license (CC-BY-4.0). You can download the raw and published (GFA/RDF/FASTA) data and store it for further processing.

11 Is this about free software?

Absolutely. Free software allows for fully reproducible pipelines. You can take our workflows and data and run it elsewhere!

12 How do I upload raw data?

We are preparing raw sequence data pipelines (fastq and BAM). The reason is that we want the best data possible for downstream analysis (including protein prediction and test development). The current approach where people publish final sequences of SARS-CoV-2 is lacking because it hides how this sequence was created. For reasons of reproducible and improved results we want/need to work with the raw sequence reads (both short reads and long reads) and take alternative assembly variations into consideration. This is all work in progress.

We also have a project going for uploading sequence data directly from the sequencer into PubSeq using the NVIDIA Jetson.

13 How do I change metadata?

14 How do I change the work flows?

Workflows are on github and can be modified. See also the BLOG on workflows.

15 How do I change the source code?

Go to our source code repositories, fork/clone the repository, change something and submit a pull request (PR). That easy! Check out how many PRs we already merged.

16 Should I choose CC-BY or CC0?

Restrictive data licenses are hampering data sharing and reproducible research. CC0 is the preferred license because it gives researchers the most freedom. Since we provide metadata there is no reason for others not to honour your work. We also provide CC-BY as an option because we know people like the attribution clause.

In all honesty: we prefer both data and software to be free.

17 Are there also variant in the RDF databases?

We do output a RDF file with the pangenome built in, and you can parse it because it has variants implicitly.

We are also writing tools to generate VCF files directly from the pangenome.

18 How do I deal with private data and privacy?

A public sequence resource is about public data. Metadata can refer to private data. You can use your own (anonymous) identifiers. We also plan to combine identifiers with clinical data stored securely at REDCap. See the relevant tracker for more information and contributing.

19 Do you have any checks or concerns if human sequence accidentally submitted to your service as part of a fastq? *

We are planning to remove reads that match the human reference.

20 Does PubSeq support only SARS-CoV-2 data?

To date, PubSeq is a resource specific to SARS-CoV-2, but we are designing it to be able to support other species in the future.

21 Contact: how do I communicate with you?

We use a matrix room you can join. See also contact.

22 Citing PubSeq

We have two publications in the works. Until we have a DOI please cite PubSeq in the following way:

We made use of the COVID-19 public sequence (PubSeq) resources hosted at

23 Who are the sponsors?

The main sponsors are listed in the footer. In addition to the time generously donated by many contributors we also acknowledge Amazon AWS for donating COVID-19 related compute time.

Edit text!