A Highly Scalable and Rapidly Deployable RNA Extraction-Free COVID-19 Assay by Quantitative Sanger Sequencing

Devon Chandler-Brown; Anna M. Bueno; Oguzhan Atay; David S. Tsao

In this report, we describe a molecular diagnostic method for COVID-19 based on quantitative Sanger (qSanger), a technique for reconstructing quantitative data from Sanger sequencing chromatograms that we originally developed and validated for non-invasive prenatal testing (NIPT) of cell-free DNA (cfDNA) [8] . The workflow of qSanger-based COVID-19 is highly similar to traditional Sanger sequencing of reverse transcription (RT)-PCR amplicons (Fig. 1 ), but it includes the addition of a frame-shifted synthetic COVID-19 spike-in DNA in the reaction master mix. qSanger COVID-19 is designed to support one-step reverse-transcription (RT)-PCR directly from viral transport media (VTM) of specimens, without an RNA purification step (Fig. 1A) . The amplification products are then purified and Sanger sequenced by automated capillary electrophoresis. Synthetic DNA included in RT-PCR master mix prior to PCR amplification serves as an internal control that enables specimens to be readily identified as either positive or negative for COVID-19 ( Fig. 1B-C) . Quantitative analysis of the Sanger sequence chromatogram gives qSanger an extremely high sensitivity and specificity for all positive results with a limit of detection of 10-20 genome copy equivalents (GCE), equivalent to gold-standard qPCR methods. Furthermore, the presence of a spike-in as an intra-sample control in the qSanger assay allows for easy interpretation of results and determination of sources of error (e.g. extraction or amplification or sequencing failure), and also allows population-level analyses such as mutational analysis and contact tracing. In addition, the ratio of the amplitudes of corresponding bases between the endogenous and spike-in sequences at offset positions reflects the ratio of the molecular abundances of the two sequences. Computationally combining the amplitude ratios of multiple corresponding bases can then be used to estimate the viral load over a 400-fold dynamic range with Poisson-limited coefficient of variation.