Back

ct2vl: Converting Ct Values to Viral Loads for SARS-CoV-2 RT-qPCR Test Results

Hill, E.; Yilmaz, F.; Callahan, C. J.; Cheng, A.; Braun, J.; Arnaout, R.

2022-06-21 microbiology
10.1101/2022.06.20.496929 bioRxiv
Show abstract

RT-qPCR is the de facto reference method for detecting the presence of SARS-CoV-2 genomic material in infected individuals (1). Although RT-qPCR is inherently quantitative and despite SARS-CoV-2 viral loads varying by 10 orders of magnitude and therefore being potentially highly clinically informative, in practice SARS-CoV-2 RT-qPCR results are usually reported qualitatively as simply positive or negative. This is both because of the mathematical complexity of converting from Ct values to viral loads and because the same Ct value can correspond to orders-of-magnitude differences in viral load depending on the testing platform (2, 3, 4). To address this problem, here we present ct2vl, a Python package designed to help individual clinical laboratories, investigators, and test developers convert from Ct values to viral loads on their own platforms, using only the data generated during validation of those platforms. It allows any user to convert Ct values to viral loads and is readily applicable to other RT-qPCR tests. ct2vl is open source, has 100% code coverage, and is freely available via the Python Package Index (PyPI). IMPORTANCEUp to now, COVID-19 test results have been reported as positive vs. negative, even though "positive" can mean anywhere from 1 copy of SARS-CoV-2 virus per milliliter of transport media to over 1 billion copies/mL, with attendant clinical consequences. Democratizing access to this quantitative data is the first step toward its eventual incorporation into test development, the research literature, and clinical care.

Matching journals

The top 9 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.