Quality of whole genome sequencing from blood versus saliva derived DNA in cardiac patients

Abstract

Background: Whole-genome sequencing (WGS) is becoming an increasingly important tool for detecting genomic variation. Blood derived DNA is the current standard for WGS for research or clinical purposes but may not always be feasible to acquire. The usability of DNA from saliva for WGS is not known. We compared the quality of WGS between blood versus saliva derived DNA. Methods: WGS was performed in DNA from 531 blood and 502 saliva samples (including 5 paired samples) from participants enrolled in a heart disease biorepository. We compared the proportion of sequencing reads that mapped to non-human sources (microbiome), the sequencing coverage, and the yield and concordance of single nucleotide variant (SNV) and copy number variant (CNV) calls between blood and saliva genomes. Results: Of 531 blood and 502 saliva samples, 46% saliva DNA failed quality control (QC) requirements for WGS compared to 6% QC failure for blood DNA. An average of 10.7% WGS reads in the saliva samples mapped to the human oral microbiome compared to 0.09% WGS reads in blood samples. However, these reads were readily excluded by excluding reads that did not map to the human reference genome. Sequencing coverage met or exceeded the target sequencing depth of 30x in all the blood samples and 4 of the 5 saliva samples; the fifth saliva sample had an average sequencing depth of 22.6x. Over 95% of SNVs identified in saliva were concordant with those identified in blood across the genome, within all gene coding regions, and within cardiovascular diseaserelated gene coding regions. Rare SNVs, defined as those with a minor allele frequency of less than 1% in the Genome Aggregation Database, had a lower concordance of 90% between blood and saliva genomes. CNVs had only 76% concordance between blood and saliva samples. Conclusions: High quality saliva samples that meet stringent QC criteria can be used for WGS when blood-derived DNA is not available or is not suitable. Saliva DNA provides an acceptable yield of SNV calls but has a lower yield for CNV calls compared to blood DNA.

Evaluation of saliva as a source of accurate whole-genome and microbiome sequencing data

“…This enabled us to confidently establish that the differences between blood and saliva were entirely similar to the differences that one should expect if either an individual’s blood or saliva had been sequenced twice. Thus, while high concordance between sequencing data in blood and saliva has previously been reported [18–20,28,38], we have been able to better contextualise such a result. This has afforded the conclusions that the WGS data generated from the saliva collection kits is of high quality; and that it will be possible to analyse the prospective POPGEN dataset alongside existing datasets (for analyses such as common or rare variant association studies) without fear of harmful batch effects.…”

“…In our study, we have concentrated on appraising the suitability of saliva for WGS in humans for the calling of SNPs and INDELs. In both Yao et al [18] and Trost et al [28] the question of the accuracy of the calling copy-number variations (CNVs) from saliva is also discussed where greater discordance (than for SNPs or INDELs) between blood and saliva has been presented. Prospectively, the 39 paired datasets of GAZEL-ADN will provide valuable opportunities for evaluating of the calling of CNVs from saliva and also for similar investigations for structural variants, transposable elements, and mitochondrial variation.…”

“…Limiting factors for saliva for sequencing were however presented in [16] where the presence of non-human DNA and subsequent low quantities of attained human DNA was postulated as resulting in large observed differences in the number of markers that could be called (significantly less in saliva than in blood); this being an observation that also made by Herráez and Stoneking [17]. Indeed, even in a recent study that compared WGS data between blood and saliva [18], the most problematic characteristics regarding data from saliva included a large number of samples failing an initial agarose gel quality control test (suggesting very poor quality DNA) and the potential for the presence of a large proportion of non-human DNA. The possibility for low quality [19] and low quantity [20] of human DNA in certain saliva samples has also been presented in the sequencing data generated via exome-capture kits.…”

Early Environmental Exposures and Contaminants: a Design Framework for Biospecimen Collection and Analysis for a Prospective National Birth Cohort

“…Whole genome sequencing (WGS) should be considered as costs continue to decline, which would allow for the analysis of rare variants, which may be particularly useful in the context of prenatal investigation (e.g., de novo mutation). Notably, while blood is the current standard for WGS, emerging evidence suggests that high-quality salivary samples with sufficient DNA yield are adequate for WGS (Yao, Akinrinade, Chaix, & Mital, 2020). Finally, while salivary samples for genomic investigation are considered essential, we suggest allowing participants to opt-out, should they not want to provide genomic data; this is especially important given concerns that some communities and individuals may hold with genomic investigation (Santos, 2008).…”

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

“…Collection of saliva DNA is commonly used for population studies and to obtain samples from remote areas 11 – 15 . The accuracy of genotyping in saliva-derived DNA has been reported as comparable to DNA derived from blood 16 – 18 .…”