Accurate Identification of SARS-CoV-2 from Viral Genome Sequences using Deep Learning

Alejandro Lopez-Rincon; Alberto Tonda; Lucero Mendoza-Maldonado; Eric Claassen; Johan Garssen; Aletta D. Kraneveld

From the repository 2019 Novel Coronavirus Resource (2019nCoVR) [6] , we 80 downloaded all the available sequences with the query Nucleotide Complete-ness="complete" AND host="homo sapiens", for a total of 588 samples. Next, we removed all repeated sequences, resulting in 553 unique sequences of variable length (1,260-31,029 bps). The data was organized and labeled as summarized by Table 1 . We grouped HCoV-229E and HCoV-OC43 in the same class, as they 85 are mostly known as Coronaviruses responsible for the common cold [27] ; the two available samples of HCoV-4408 were also added to the same class, as it is a Betacoronavirus 1, as HCoV-OC43. In a similar fashion, we grouped HCoV-NL63 and HCoV-HKU1, as they are both associated with acute respiratory 5 . CC-BY-NC 4.0 International license author/funder. It is made available under a The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.03.13.990242 doi: bioRxiv preprint infections (ARI) [28] . Finally, we grouped SARS-CoV/SARS-CoV-P2/SARS-