Efficient wastewater sample filtration improves the detection of SARS-CoV-2 variants: An extensive analysis based on sequencing parameters

Abstract

During the SARS-CoV-2 pandemic, many countries established wastewater (WW) surveillance to objectively monitor the level of infection within the population. As new variants continue to emerge, it has become clear that WW surveillance is an essential tool for the early detection of variants. The EU Commission published a recommendation suggesting an approach to establish surveillance of SARS-CoV-2 and its variants in WW, besides specifying the methodology for WW concentration and RNA extraction. Therefore, different groups have approached the issue with different strategies, mainly focusing on WW concentration methods, but only a few groups highlighted the importance of prefiltering WW samples and/or purification of RNA samples. Aiming to obtain high-quality sequencing data allowing variants detection, we compared four experimental conditions generated from the treatment of: i) WW samples by WW filtration and ii) the extracted RNA by DNase treatment, purification and concentration of the extracted RNA. To evaluate the best condition, the results were assessed by focusing on several sequencing parameters, as the outcome of SARS-CoV-2 sequencing from WW is crucial for variant detection. Overall, the best sequencing result was obtained by filtering the WW sample. Moreover, the present study provides an overview of some sequencing parameters to consider when optimizing a method for monitoring SARS-CoV-2 variants from WW samples, which can also be applied to any sample preparation methodology.

Fig 1. Effect of filtration (F) and purification treatment (T) on the number of SARS-CoV-2 viral copies used as input for the library preparation.

Copy number yield for the N1 gene was obtained by RT-qPCR from three biological samples processed according to four experimental conditions: F-NT: WW filtration ‐ no RNA treatment; F-T: WW filtration ‐ RNA treatment; NF-NT: no WW filtration ‐ no RNA treatment; NF-T: no WW filtration ‐ RNA treatment. Two-way ANOVA (p ≤ 0.05). Bar plots show mean and standard deviation and asterisks show significance level upon t-test (not significant (no asterisks) = p > 0.05; * = p ≤ 0.05; ** = p ≤ 0.01; *** = p ≤ 0.001; **** = p ≤ 0.0001, n = 3).

Fig 2. Library quality assessment using the Agilent 2100 Bioanalyzer System-High Sensitivity DNA Kit.

A) Data displayed as a gel-like image. The upper (purple) and the lower (green) bands represent the internal markers. B) Electropherogram traces represent the expected size distribution of the library fragments at 75 seconds (350–400 bp). The color of the line indicates the different conditions: F-NT (green), F-T (red), NF-NT (light blue) and NF-T (blue). The left (approximately 45 seconds) and far-right peaks (approximately 115 seconds) are internal markers.

Fig 3. Taxonomic classification of reads using the Dragen metagenomics pipeline.

A) Percentage of reads classified as belonging to SARS-CoV-2 (Two-way ANOVA (p ≤ 0.05)); B) Percentage of reads classified as belonging to Bacteria (Two-way ANOVA (p ≤ 0.01)); Bar plots show mean and standard deviation and asterisks show significance level upon t-test (not significant (no asterisks) = p > 0.05; * = p ≤ 0.05; ** = p ≤ 0.01; *** = p ≤ 0.001; **** = p ≤ 0.0001, n = 3).

Fig 4. Percentage of the SARS-CoV-2 genome covered at least 30X.

Percentage of the SARS-CoV-2 genome which is covered at least 30 times by the sequencing output reads. Each condition is the mean of three biological replicates (Two-way ANOVA (p ≤ 0.05)). Bar plots represent mean and standard deviation, and asterisks indicate significance level upon t-test (no asterisks = p > 0.05; * = p ≤ 0.05; ** = p ≤ 0.01; *** = p ≤ 0.001; **** = p ≤ 0.0001, n = 3).

Fig 5. Frequency of nt and AA mutations with a mutation frequency of ≤ 50%, ≤ 10% or ≤ 5% and genome coverage ≥30X.

A) Frequency of nucleotide mutations with a cut-off mutation frequency of ≤ 50%, ≤ 10% or ≤ 5% and genome coverage ≥30X (Two-way ANOVA (p ≤ 0.001)). B) Frequency of amino acid mutations with a cut-off mutation frequency of ≤ 50%, ≤ 10% or ≤ 5% and coverage ≥30X (Two-way ANOVA (p ≤ 0.001)). Bar plots show mean and standard deviation and asterisks show significance level upon t-test (no asterisks = p > 0.05; * = p ≤ 0.05; ** = p ≤ 0.01; *** = p ≤ 0.001; **** = p ≤ 0.0001, n = 3).

Fig 6. Omicron BA.1 nt and AA spike gene mutations.

The heatmap shows the presence of spike gene mutations associated with the Omicron BA.1 variant in all the experimental samples analyzed. Each row represents one mutation. The number inside the bullet represents the mutation frequency. The total number of mutations detected is shown at the bottom of each column. Black circles represent mutations with less than 30X coverage. The heatmap was generated with Morpheus, https://software.broadinstitute.org/morpheus.

Fig 7. Summary heatmap of the evaluated post-sequencing parameters.

The summary heatmap shows the results obtained from the analysis of the different sequencing parameters for each experimental condition (top right columns: F-NT, F-T, NF-NT, NF-T). Dark blue color represents 0%, red color represents 100% while the white color represents 50%. The color scale is relative to the highest and lowest value for each parameter among the treatments.