Further analysis of metagenomic datasets containing GD and GX pangolin CoVs indicates widespread contamination, undermining pangolin host attribution
The only animals other than bats reported to have been infected with SARS-CoV-2-related coronaviruses (SARS2r-CoVs) prior to the COVID-19 pandemic are pangolins. In early 2020 multiple papers reported the identification of two clades of SARS2r-CoVs, GD and GX, infecting pangolins. However the RNA-Seq datasets supporting pangolin genome assembly were widely contaminated, contained synthetic vectors or were heavily enriched or filtered with little but coronavirus sequences left in the datasets. Here we investigate two pangolin fecal samples sequenced by Li et al. (2021) provided in support of GD PCoV infection of pangolins in Guangdong and find the read distribution consistent with PCR amplicon contamination and SARS-CoV-2 contamination, and further identify the presence of synthetic plasmid sequences. We also build upon our previous work to further analyze the dataset GX/P3B by Lam et al. (2020), which is the only non enriched/heavily filtered pangolin tissue dataset sequenced by Lam et al. (2020). We identify synthetic vectors and confirm human genomic origin samples in the dataset. Finally, we find human mitochondrial sequences in all pangolin organ datasets and mouse and tiger mitochondrial sequences in selected pangolin organ datasets sequenced by Liu et al. (2019). We infer that human and mouse genomic origin sequences were probably sourced from contamination prior to sequencing, while tiger origin sequence contamination may have occurred due to index hopping during sequencing. These observations are problematic for attributing pangolins as SARS2r-CoV hosts in the datasets examined. The forensic methods developed and used here can be applied to examine any third party SRA data sets.