SwabSeq TLDR

This document shares some aspects of Octant's platform that we believe would enable scaling of COVID19 sequence-based testing. The protocol, which we call SwabSeq, allows for detection of SARS-COV-2 genomic RNA, without purification, in a single-step RT-PCR followed by sequencing. This eliminates some of the current bottlenecks to most current COVID sequencing protocols (e.g. rare/expensive reagents, extra purification steps, qPCR), and also utilizes multiplexing to sequence larger sets of samples without extensive automation.

SwabSeq is:

This method does take time (~12 hours) and requires specialized equipment such as PCR machines and a sequencer, and thus is best suited for centralized testing facilities where scale can be achieved. In addition, there are still technical questions left to be answered, but we feel the results are promising enough for some of our collaborators to start testing on clinical samples. Finally, new technical methodologies such as this will not solve any of the considerable logistical problems of swabbing, data entry, sample processing into microtiter plates, and returning results to that many people at once.


At Octant, we have built, optimized and are running at scale a high-throughput RNA-amplicon sequencing platform in lysates of human cell cultures. In particular, we've optimized this process to be sensitive to low numbers of RNA molecules, to be low touch, and cost-effective. To give some idea, the process we run at Octant, which is more involved and expensive than the process described here, runs efficiently at scales up to 10,000 samples at a time and is conducted by a single person with little to no automation and feeds into automated informatic pipelines.

Here we describe our efforts over the last two weeks to repurpose this platform for COVID detection. We decided to do this because the methodology allows to get over certain bottlenecks in the testing processing including RNA purification, qPCR machinery and automation. This does not solve the logistics of swabbing and tracking tens of thousands to hundreds of thousands of people a day, which is likely another large limitation. However, there is little impetus to solve the logistical problems if there are downstream technological limitations on conducting the actual tests at that scale.

We provide a set of protocols, software, notebooks, primer designs, and proof-of-concept experiments to establish limits of detection and reliability of an NGS-based amplicon sequencing protocol. This is a living website, and we will continue to update this as we analyze the datasets and learn more through our own work and that of our collaborators.

We are opening this up now, because we are sufficiently confident that this can help a number of groups trying to scale testing, and we are releasing far before pre-print because we think urgency is of the utmost importance. We have applied for patent rights in some aspects of these protocols, but are freely licensing them for use against COVID using the Open COVID License terms linked here.

Summary of SwabSeq

SwabSeq is a simple and scalable protocol light on reagent and automation requirements.

  1. Put swab directly into lysis buffer and inactivate (we currently use this).
  2. We add this inactivated lysate directly into a one-step RT-PCR mix, along with barcoded primers unique to each well, and a synthetic RNA spike-in control
    1. We include a synthetic RNA spike-in that contains same priming regions as the target (COVID sequence) we are trying to detect, but the amplicon has a unique string of nucleotides that let's us differentiate it from actual virus by sequencing. This serves as a well control for reverse transcription and amplification of the virus sequence, but also as a way to allow for amplification even in negative samples and thus providing a normalization between wells.
    2. In addition to the spike-in and virus, we reverse transcribe and amplify a human housekeeping gene to control for the swabbing itself.
    3. The primers all have p5 & p7 sequences and indices for the Illumina platform already
    4. We have tested a number of RT's over years; there are many that work equivalently.
  3. We pool and sequence