Contaminating DNA in human saliva alters the detection of variants from whole genome sequencing

Abstract

Cells obtained from human saliva are commonly used as an alternative DNA source when blood is difficult or less convenient to collect. Although DNA extracted from saliva is considered to be of comparable quality to that derived from blood, recent studies have shown that non-human contaminating DNA derived from saliva can confound whole genome sequencing results. The most concerning complication is that non-human reads align to the human reference genome using standard methodology, which can critically affect the resulting variant genotypes identified in a genome. We identified clusters of anomalous variants in saliva DNA derived reads which aligned in an atypical manner. These reads had only short regions of identity to the human reference sequence, flanked by soft clipped sequence. Sequence comparisons of atypically aligning reads from eight human saliva-derived samples to RefSeq genomes revealed the majority to be of bacterial origin (63.46%). To partition the non-human reads during the alignment step, a decoy of the most prevalent bacterial genome sequences was designed and utilised. This reduced the number of atypically aligning reads when trialled on the eight saliva-derived samples by 44% and most importantly prevented the associated anomalous genotype calls. Saliva derived DNA is often contaminated by DNA from other species. This can lead to non-human reads aligning to the human reference genome using current alignment best-practices, impacting variant identification. This problem can be diminished by using a bacterial decoy in the alignment process.

The human “contaminome”: bacterial, viral, and computational contamination in whole genome sequences from 1000 families

“…In both genomic analysis of a single organism and metagenomics studies, such contamination can have critical impacts on downstream analysis. In whole-genome sequencing studies, bacterial contamination can result in false alignments and erroneous downstream variant calls 16 , 17 . In studies of the microbiome, contamination can distort the estimation of microbial abundance of different genera 13 , 18 .…”

The Human “Contaminome”: Bacterial, Viral, and Computational Contamination in Whole Genome Sequences from 1,000 Families

“…In both genomic analysis of a single organism and metagenomics studies, such contamination can have critical impacts on downstream analysis. In whole-genome sequencing studies, bacterial contamination can result in false alignments and erroneous downstream variant calls [17, 18]. In studies of the microbiome, contamination can distort the estimation of microbial abundance of different genera [14, 19].…”

DNA Methylation Variation Is Identified in Monozygotic Twins Discordant for Non-syndromic Cleft Lip and Palate

“…Salivary glands, lip, and palate all arise from neural crest cells, making cells from saliva a developmentally appropriate proxy for these tissues. In addition, saliva is a good source of high quality DNA for use in epigenomics ( Thompson et al, 2013 ; Smith et al, 2015 ) and has been deemed more informative than blood for DNA methylation analysis ( Lowe et al, 2013 ) in spite of the possibility of bacterial contamination resulting in non-human reads in human sequencing data ( Samson et al, 2020 ).…”