Comparative assessment of Nanotrap and polyethylene glycol-based virus concentration in wastewater samples

Abstract Wastewater-based epidemiology is now widely used in many countries for the routine monitoring of SARS-CoV-2 and other viruses at a community level. However, efficient sample processing technologies are still under investigation. In this study, we compared the performance of the novel Nanotrap® Microbiome Particles (NMP) concentration method to the commonly used polyethylene glycol (PEG) precipitation method for concentrating viruses from wastewater and their subsequent quantification and sequencing. For this, we first spiked wastewater with SARS-CoV-2, influenza and measles viruses and norovirus and found that the NMP method recovered 0.4%–21% of them depending on virus type, providing consistent and reproducible results. Using the NMP and PEG methods, we monitored SARS-CoV-2, influenza A and B viruses, RSV, enteroviruses and norovirus GI and GII and crAssphage in wastewater using quantitative PCR (qPCR)-based methods and next-generation sequencing. Good viral recoveries were observed for highly abundant viruses using both methods; however, PEG precipitation was more successful in the recovery of low-abundance viruses present in wastewater. Furthermore, samples processed with PEG precipitation were more successfully sequenced for SARS-CoV-2 than those processed with the NMP method. Virus recoveries were enhanced by high sample volumes when PEG precipitation was applied. Overall, our results suggest that the NMP concentration method is a rapid and easy virus concentration method for viral targets that are abundant in wastewater, whereas PEG precipitation may be more suited to the recovery and analysis of low-abundance viruses and for next generation sequencing.


Introduction
Wastewater-based epidemiology (WBE) is an important asset used for providing public health insights for the monitoring of infectious diseases at a community level.Many pathogens, including enteric and r espir atory viruses, ar e excr eted in the faeces and urine of infected individuals and, hence, can be isolated from municipal w astew ater.The vir al concentr ation dynamics in se wa ge can indicate the r elativ e abundance of cases within a community (Jiang et al. 2022, Reynolds et al. 2022 ).WBE has been a valuable auxiliary surveillance tool for those pathogens that are associated with asymptomatic cases.For instance, WBE has been used for the comm unity-le v el surv eillance of poliovirus for decades (Pa vlo v et al. 2005, Rakoto-Andrianarivelo et al. 2008, Hovi et al. 2012, O'Reill y et al. 2018, Kla psa et al. 2022 ).Since the start of the COVID-19 pandemic in 2020, many countries have utilised WBE for quantitative tracking, early warning and variant-level monitoring for SARS-CoV-2 (Ai et al. 2021, Carcer en y et al. 2021, Kumar et al. 2021, Wang et al. 2022 ).
In most w astew ater surv eillance pr ogr ammes, se wa ge samples are taken daily or multiple times a week and tr ansferr ed to a laboratory for analysis.To utilise the WBE approach successfull y, viruses typicall y need to be concentrated in the samples to enable their detection and quantification.This concentration step can be performed using a range of methods including electr onegativ e/electr opositiv e filtr ation, ultr afiltr ation, ultr acentrifugation or precipitation with polyethylene glycol (PEG), ammonium sulphate or skimmed milk (Farkas et al. 2020a, Ahmed et al. 2020, Rusiñol et al. 2020, Philo et al. 2021, K e vill et al. 2022 ).It is important that the concentration method successfully recovers the viruses from the samples while eliminating any impurities that may adv ersel y affect downstr eam pr ocesses, suc h as nucleic acid extr action, vir al detection, and quantification (Ahmed et al. 2022a ).In most studies, either quantitative or digital PCR (qPCR or dPCR) are used to quantify the viral genomes (Corpuz et al. 2020 ;Farkas et al. 2020bFarkas et al. , 2020c ) ) and, when sequencing is utilised, this often also r equir es amplification of the target viruses (Karthik e yan et al. 2022 ).While PCR-based a ppr oac hes enable the r a pid, sensitiv e and when needed, str ain-le v el detection of the target, they may be affected by residual organic matter that can interfere with the reverse transcriptase and DNA polymerase enzymes (Ahmed et al. 2022a ).Ho w e v er, while dPCR is less affected by such inhibitors (Ahmed et al. 2022b, Jahne et al. 2020, Flood et al. 2021 ), the equipment r equir ed is not available in many WBE laboratories .T he viral concentration method should therefore aim to reduce the concentration of organic matter to ensure high quality r esults, r egardless of the detection method used.
Recentl y, the a pplication of ma gnetic beads to concentrate viruses in w astew ater has been suggested, but only a few studies ar e av ailable on the de v elopment and use of this tec hnique (Karthik e yan et al. 2021, Ahmed et al. 2023, Andersen et al. 2023, Daza-Torres et al. 2023, Feng et al. 2023 ).The method is quic kl y and easily performed without the requirement for complex lab equipment and, hence, may be applied for on-site analysis that facilitates the delivery of rapid insights.It is therefore important to impr ov e our understanding of the ov er all performance of suc h methods.
In this study, we explored the usefulness of the Nanotrap ® Micr obiome P articles (NMP, formerl y called Nanotr a p ® Ma gnetic Virus Particles), for the recovery of different human pathogenic viruses and a faecal indicator virus fr om waste water.First, we used samples spiked with the target viruses to estimate recovery efficienc y.Then, w e performed an intr a-labor atory trial using ma gnetic bead concentr ation along with PEG pr ecipitation to explor e r epr oducibility.Lastl y, we trialled the effect of the magnetic bead concentration method along with PEG precipitation on 42 w astew ater samples to investigate sensitivity.

Virus spiking
In order to test the feasibility of the NMP method for virus r ecov ery, ion-exc hanged water and wastewater samples were spiked with known concentrations of heat-inactivated SARS-CoV-2 (kindly provided by Prof Richard Stanton, Cardiff University), influenza A/California/07/2009 (H1N1), B/Lee/40 (kindly provided by Dr Eleanor Gaunt, University of Edinbur gh), nor ovirus GII (NoVGII) in diluted and filtered faecal matter from a patient with confirmed infection (kindly provided by Dr Lydia Drumwright, University of Cambridge), measles virus (MeV) in the form of a vaccine (VWR International, USA) and Phi6 bacteriopha ge, cultur ed in-house (K e vill et al. 2022 ).Samples wer e pr ocessed in triplicate.

Intr a-labor a tory assessment
To test the r epr oducibility of the NMP concentration and the PEG precipitation methods, four experienced lab staff members received the same w astew ater sample , in triplicate , for each of the methods (NMP, PEG-150, PEG-37.5),resulting in a total of nine identical samples per each of the four individuals .T he unspiked w astew ater samples, which were processed as detailed below.

Waste wa ter samples
For the spiking and intr a-labor atory trial, 20 l and 5 l influent w astew ater samples w ere collected using grab sampling at the Bangor w astew ater treatment plant (Bangor, Wales) on the 5 th Nov 2021 and on the 18 th August 2022, r espectiv el y.These waste water samples contained negligible amounts of the target viruses.The samples were processed in triplicate.Further 42 composite w astew ater samples w ere collected, as part of the Welsh Na-tional Wastewater Monitoring programme between 28 th and 30 th No vember 2022.T hese samples wer e pr ocessed without r eplication.The pH, turbidity, electrical conductivity, ammonium and orthophosphate concentrations of the samples wer e measur ed as described pr e viousl y (Hillary et al. 2021, Farkas et al. 2022 ).

PEG precipitation
All samples, except those spiked with human viruses, were concentrated using PEG precipitation, as described previously (Farkas et al. 2021 ).In brief, 200 ml and 50 ml w astew ater samples w ere centrifuged at 10 000 × g at 4 • C for 10 min and then 150 ml or 37.5 ml of the resulting supernatant was spiked with known quantities of Phi6 bacteriophage as a process control virus.After pH adjustment to 7-7.5, the solution was mixed with PEG 8000 and NaCl to a final concentration of 10% and 2%, r espectiv el y.After a 16 h incubation at 4 • C, the samples were centrifuged at 10 000 × g at 4 • C for 30 min and the resulting pellet was subject to nucleic acid extraction.We refer to the PEG method used on high volume (150 ml supernatant) samples as the PEG-150 method, and we use the term PEG-37.5 when the method was applied on low volume (37.5 ml supernatant) samples.

Nanotrap ® Microbiome Particles (NMP) concentr a tion method
The NMP kit was obtained from Ceres Nanoscience Inc., Manassas, VA, USA.At the time of purchase, the product was named Nanotr a p ® Ma gnetic Virus Particles and supplied with Nanotrap ® Enhancement Reagent 2 (ER2).The kit was used as per the manufacturer's instructions .T he samples were centrifuged and spiked, when applicable, as described above.Then, 400 μl of the ER2 buffer was added to 45 ml sample supernatant and vortexed to mix, follo w ed b y the addition of 600 μl Nanotr a p beads.Samples wer e inverted to mix and incubated at room temperature for 10 min, which included an inversion at the 5-minute mark.Tubes containing beads were then placed onto a magnetic rack and once the solution became clear with the beads adhered to the side of the tube, the solution was then r emov ed.The beads were recovered in 1 ml molecular-grade w ater follo w ed b y ma gnetic separ ation and the r emov al of the solution.The r ecov er ed beads wer e then subject to nucleic acid extraction.

RN A/DN A extr action
Viral nucleic acids were recovered from PEG pellets or from NMP concentrate using the NucliSens extraction system (BioMerieux, France) on a KingFisher automated extraction system (Thermo Fisher, USA) as described pr eciousl y (K e vill et al. 2022 ).In brief, the pellets or beads were resuspended in 850 μl Lysis Buffer, mixed and incubated for at least 10 min follo w ed b y the addition of the NucliSens magnetic silica beads for DN A/RN A binding.The beads w ere then w ashed with NucliSens Wash Buffer #1 and #2 twice and with Wash Buffer #3 once .T he nucleic acids were then eluted from the beads in Wash Buffer #3 at 60 • C. The final volume of the eluate was 100 μl.

Virus quantification
The target RNA viruses were quantified using RT-qPCR on a QuantStudio Flex 6 system (Applied Biosystems, USA) as described pr e viousl y (Farkas et al. 2022 ).The SARS-CoV-2 N1 gene fr a gment and phi6 bacteriophage, andthe influenza A and B viruses (FluA and FluB) were assayed in two duplexed qPCR reactions using validated primers and probes (Gendron et al. 2010, CDC 2020, Shu et al. 2021 ).Enterovirus spp.(EV) and norovirus GI and GII (NoVGI and NoVGII) were quantified using a triplex assay while measles virus (MeV) was quantified with a singleplex assay with validated primers and pr obes (Gr egory et al. 2006, Hummel et al. 2006, ISO/TS 2019 ).In brief, the reaction mixes for RNA viruses contained TaqMan viral 1-step RT-qPCR master mix (Applied Biosystems, Inc., USA), 1 μg bovine serum albumin (BSA), 10 μM forw ar d, 20 μM r e v erse primers and 5 μM probe.For the duplex SARS-CoV-2/Phi6 and FluA/FluB assa ys , 16 nmol MgSO 4 was also added.The amplification was carried out using the following conditions: r e v erse tr anscription at 50 • C for 30 min follo w ed b y enzyme inactivation at 95 • C for 20 s, then 45 amplification cycles of 95 • C for 3 s, 60 • C for 30 s.
CrAssphage qPCR was set up using the QuantiNova lo w-R O X pr obe qPCR mix (Qia gen, German y), 1 μg bovine serum albumin (BSA), 10 μM forw ar d and r e v erse primers and 5 μM probe (Stachler et al. 2018, Farkas et al. 2022 ).The reaction conditions were as follo ws: DN A denaturation at 95 • C for 2 min follo w ed b y 40 c ycles of 95 • C for 15 s and 60 • C for 1 min.
All samples were run in duplicate and quantification was carried out using a 10 5 -10 0 genome copies (gc)/ μl dilution series of synthetic RN A oligo standar ds (SARS-CoV -2 and phi6), commercial genomic standards (FluA/B; Twist Bioscience, USA), RNA extr acted fr om MMR v accine (MeV) or plasmid DNA (NoVGII and crAsspha ge).Eac h plate contained multiple non-template controls to assess cross-contamination.

SARS-CoV-2 sequencing
A subset of SARS-CoV-2 samples processed by the two methods were sequenced ( Table S1 ) to compare the quality of RNA template for variant detection.Follo wing extraction, RN A w as purified using a standardised protocol with magnetic bead clean-up of 1.8X Mag-Bind Total NGS beads (Omega BioTek).A LunaScript RT Supermix Kit (New England Biolabs, UK) was then used to synthesise cDNA before sequencing libraries were prepared using NimaGen's EasySeq RC-PCR SARS-CoV-2 whole genome sequencing kit (Nimagen, The Netherlands).The pooled library was spiked with a control (an adapter ligated library supplied by Illumina Inc., San Diego, CA) and run on an Illumina NextSeq 1000 system using a P1 kit (2 ×150 bp) following concentration loading guidelines provided by Illumina.

Da ta anal yses
Initial data analysis and quality control for the qPCR data were performed using the QuantStudio Real-time PCR software v1.7 (Applied Biosystems, USA), following MIQE Guidelines (Bustin et al. 2009 ), with slope between -3.6 and -3.1, efficienc y betw een 90% and 100%.The LOD and LOQ of target viruses has pr e viousl y been published (Farkas et al. 2022 ).Sample concentrations were expressed as gc/ μl nucleic acid extract.Virus concentrations were transformed to gc/l as follows: The full dataset is displayed in Table S1 .
The data from duplicate reactions were combined and the avera ge v alue was used for statistical anal ysis.Sha pir o-Wilk test confirmed that the data were non-normally distributed ( P < 0.001).The difference among users and methods performance was assessed using Mann-Whitney U test and Kruskal-Wallis tests.Spearman's r ank corr elation was used to assess the correlation between viral concentrations , reco veries and w astew ater physicoc hemical pr operties.Statistical anal yses wer e performed using SPSS v27 (IBM Inc., USA).
We then estimated r elativ e abundance of SARS-CoV-2 lineages of mixed-lineage virus samples in w astew ater.The sequencing data were processed using Freyja v1.2.1 (Karthikeyan et al. 2022 ), which uses Single Nucleotide Variant (SNV) frequency estimation and a depth-weighted demixing tool.Sequencing data quality control (QC) pass rate was determined using the Nextflow implementation of the ARTIC pipeline ( https:// github.com/connor-lab/ ncov2019-artic-nf); a pass is ac hie v ed when > 50% of the r efer ence sequence (Genbank accession MN908947.3)bases are detected in > 10 reads.

Spiking experiment
When using the NMP concentration method, significantly higher vir al r ecov eries wer e obtained fr om spiked waste water than spiked ion-exchanged water (Mann Whitney U test; u 33, z-score 6.860, P < 0.001, Fig. 1 ).The pairwise comparison of each virus also gave similar results .T he % reco very from w astew ater using the NMP method ranged between 0.4 to 21%; the mean recovery was 6% and the median r ecov ery was 4.6%, while for ionexchanged water the precent recovery range was 0.01 to 1% with a mean r ecov ery of 0.24% and median r ecov ery of 0.12% (Fig. 1 ).Yields of spiked viruses r ecov er ed fr om waste water followed the trend: MeV (15.3%) > FluA (11.5%) > SARS-CoV-2 (7.7%) > NoVGII (1.7%) > FluB (1.3%); 10-fold higher on av er a ge than yields recover ed fr om ion-exc hanged water.

Intr a-labor a tory assessment of virus concentr a tion methods
To assess variability among lab users, four group members processed the same sample in triplicate using the NMP method (45 ml/sample), the PEG-150 (150 ml/sample), and the PEG-37.5 (37.5 ml/sample) precipitation methods .T he processed samples were tested for SARS-CoV-2, NoVGI and NoVGII, and crAssphage (FluA/B were tested for but not detected).No significant difference was found in the viral recoveries obtained by the users (Kruskal-Wallis test, P > 0.05), although User 3 slightly outperformed the others (Fig. 2 ).
In the same experiment, significant differ ences wer e found between the performance of the methods applied for concentration of the different viruses.For SARS-CoV-2, NoVGI and NoVGII, the PEG methods outperformed the NMP method (Kruskal-Wallis test, P < 0.001).For crAssphage, the concentrations were significantly lo w er when the NMP method ( P < 0.001) and the PEG-37.5 method ( P = 0.037) were used, compared to the PEG-150 method (Fig. 2 B).

Waste wa ter testing
For further validation, 42 unspiked raw w astew ater samples w ere processed using the NMP and PEG methods and tested for a range of viruses.FluB virus was not detected in any of the samples with any of the methods, and RSV was only detected in three samples when the PEG-150 method was used (Table 1 ).The FluA virus detection rates were also substantially higher with the PEG-150 method (62%) compared with those obtained using the NMP method (26%) or the PEG-37.5 method (9%).EV sho w ed the highest detection rates when the PEG-37.5 method was used (26%), whereas the detection rates with the other two methods were lo w er .SARS-CoV -2 and NoVGI was detected in all samples when the PEG-150 method was applied, and in the majority of the samples using the other methods .NoVGII, crAssphage , and Phi6 were detected in any of the samples regardless of the method used.
Significantly higher virus concentrations were detected for Phi6 when using the NMP method, compared to the PEG-150 and PEG-37.5 methods.No other significant differences were found between methods (Fig. 3 ).
Ov er all, the vir al concentr ations and r ecov eries obtained by the different methods correlated well for SARS-CoV-2, NoV GI, NoV GII, crAssphage and Phi6 ( Table S2 ).The lack of correlation for EV and FluA may be due to the low detection rates and low virus concentr ations.Detection/r ecov ery of SARS-CoV-2, NoVGI, and NoVGII with the different methods also correlated with that of the faecal indicator virus crAssphage.Sample pH sho w ed moder ate positiv e corr elation onl y with NoVGII detection/r ecov ery when low volume of sample was PEG precipitated and a negativ e corr elation w as found betw een pH and crAssphage concentrations, detected when the sample was concentrated using the NMP method.Sample turbidity , electrical conductivity , ammonium and orthophosphate le v els sho w ed significant correlations with the human-associated virus concentr ations, especiall y when the PEG-150 method was a pplied, wher eas the viral concentr ates deriv ed fr om the other two a ppr oac hes mainl y corr elated with ammonium and orthophosphate le v els ( Table S2 ).Interestingl y, the Phi6 r ecov ery sho w ed a negativ e corr elation with sample turbidity , electrical conductivity , ammonium and orthophosphate le v els.

SARS-CoV-2 sequencing
Only 14% of samples passed QC for sequencing, comprising 11% processed with the PEG-150 method, and 3% processed using the PEG-37.5 method.All samples processed with NMP magnetic beads failed QC.Low ma pping r ates of reads to the reference genome meant that the QC threshold was difficult to pass; the percentage of bases that mapped to the SARS-CoV-2 reference genome in samples processed with the NMP beads or the PEG-37.5 method were substantially less than those processed with the PEG-150 method (1.5% vs 3.5%, r espectiv el y).Similarl y, the av era ge cov er a ge of genome depth was observ ed for the samples pr ocessed using the PEG-37.5 and the NMP methods (144X and 143X, r espectiv el y), was substantiall y less than the 248X av er a ge cov erage for samples concentrated using the PEG-150 precipitation.
Estimated linea ge pr oportions pr oduced b y F reyja w ere converted to represent the number of samples that successfully detected the lineage when processed with a specific method.Variants were detected across a higher number of sites when samples wer e pr ocessed with either of the PEG pr ecipitation methods (Fig. 4 ).Samples processed with NMP beads detected variants in the fewest number of samples for all but two variants.One sample processed with PEG failed in library preparations and so was not included in the anal yses, ther efor e v ariants detected ar e shown as a proportion of samples successfully processed.General patterns ar e congruent acr oss methods, suc h as v ariant linea ge BA.2.75 and BA.5 being the least abundant.Howe v er, a ppl ying PEG pr ecipitation with high sample volumes detected lineage BQ in 74.1% of cases making it the dominant lineage when quantifying with this method.The other methods detected BA.5 as the dominant lineage.

Discussion
In this study, we investigated the effectiveness of a novel NMP concentration method versus traditional PEG precipitation for the r ecov ery of viruses from untreated w astew ater.Overall, both appr oac hes wer e suitable for concentr ating viruses within waste water for WBE applications.We found that the NMP concentration method performed better when applied to w astew ater than for deionised water (Fig. 1 ); howe v er, man y other virus concentration methods, suc h as pr ecipitation and ultr afiltr ation a ppr oac hes, give higher recoveries in clean water than in w astew ater (Farkas et al. 2022 ).T he high NMP reco veries achieved in w astew ater may be due to the presence of ions enhancing viral binding to magnetic beads (e.g.viral aggregation, cation bridging).The viral recovery per centiles w er e compar able with those deriv ed fr om PEG pr ecipitation and ultr afiltr ation concentr ation methods, as determined pr e viousl y (Farkas et al. 2022 ) and slightly lo w er for SARS-CoV-2 compared to a similar study using NMP (Brighton et al. 2024 ).
We performed an intr a-labor atory trial during which wellmixed w astew ater subsamples w er e giv en to four experienced labor atory tec hnicians to be pr ocessed using PEG pr ecipitation methods with high and low sample volumes and using the NMP concentration method.The further steps of sample processing, nucleic acid extraction, and qPCR, were performed in batches by one person using automated systems to reduce human error.Ther efor e, onl y the sample concentration step performance was reliant on the different users.No significant differences in users' performance were noted suggesting that both methods are reproducible (Fig. 2 ).Some differences amongst users were observed when the PEG-150 method was applied, in which the final viral pr ecipitate is r esuspended by elution fr om the wall of a 250-ml centrifuge bottle.As the viral pellet is often not visible and/or spread on a wide area of the wall of the bottle, it may be hard to completely retrieve; hence, viral recovery may vary.When the PEG-37.5 method is applied, the final pellet is produced in a 50- Table 1.Detection rates (n) for SARS-CoV-2, enteroviruses (EV), norovirus GI and GII (NoV GI, NoV GII), influenza A virus (Flu A), r espir atory syncytial virus (RSV), crAssphage, and Phi6 phage process control virus in w astew ater samples using the NMP and the PEG methods for concentration.ml centrifuge tube on a small area of the tube wall, which helps ac hie v e full elution.The NMP method is less reliant on the users' skills, which is consistent with viral recoveries showing little variation in that element of the intr a-labor atory trial.We processed samples in triplicate for each method per user.This generated enough data to assess the useability of the methods, and the limitations of each method has been highlighted.In the comparison of 42 w astew ater samples processed with PEG and NMP methods sim ultaneousl y the qPCR r esults suggested that the PEG methods r ecov er ed less abundant (i.e.viruses not detected in the majority of w astew ater samples in the study) viruses (EV, Flu-A, RSV) more efficiently than the NMP method (Table 1 ).Furthermore, RSV was onl y r ecov er ed using the PEG method with high initial sample vol-umes .T he viral concentrations for SARS-CoV-2, crAssphage and noroviruses obtained by the different methods correlated well, suggesting that both NMP and PEG methods performed similarly for highly abundant viruses.Interestingly, only the process control virus Phi6 was r ecov er ed at significantl y higher concentr ations with the NMP method compared to the PEG methods, while the quantification of other viruses was not method-dependent (Fig. 3 ).The Phi6 virus used in this study was derived from an in vitro cultur ed stoc k (K e vill et al. 2022 ), whic h ma y beha v e differ entl y from viruses that are abundant in w astew ater, as viruses present in w astew ater ma y also bind to the suspended solids .

Method
In most cases, viral concentrations correlated with turbidity and chemical water parameters, suggesting that the more con- Figur e 4. T he number of samples at which SARS-CoV-2 lineages were detected using different processing methods.Maximum number of samples calculated as number of samples multiplied by number of variants: 216 samples for PEG-37.5 and NMP, 208 samples for PEG-150.Results include samples that did not pass quality c hec k. centr ated the waste water is, the mor e human viruses can be r ecov er ed.Ho w e v er, pr e vious r esearc h found that high turbidity has a negative effect on the efficiency of NMP concentration for pepper mild mottle virus (Ahmed et al. 2023 ), further suggesting that different viruses were recovered at different yields using this method.
We noted similar patterns in the sequencing data for SARS-CoV-2, with PEG-150 being able to detect less abundant variants a greater number of times .T he PEG-37.5 precipitation and NMP concentration methods failed to detect several variants in a large proportion of sites, which was correlated to the low average cover a ge and the lo w per centa ge of r eads that wer e ma pped.It is unusual to have low cov er a ge acr oss all methods; all samples were sequenced together on one run to allow comparisons to be made between pr ocessing methods, r egardless of run chemistry.How-e v er, separ ating runs based on methods would not only have given better cov er a ge (incr easing the c hance of samples passing the QC threshold) but would also have removed the possibility that a processing method may carry inhibitors (e.g.chemicals or organic matter) that affect the ov er all run.While all methods were able to distinguish between the different SARS-CoV-2 variants in samples, PEG-150 pr ecipitation demonstr ated the gr eatest ability to do this r epeatedl y and at low virus abundance, While more optimisation is still needed to impr ov e the success r ate, the incr eased sensitivity for detecting variants demonstrates the potential of this a ppr oac h.

Conclusions
Ov er all, our r esults show that the NMP method is suitable for certain WBE applications.For example, in situations where the speed of r esults (r a pid need for determination of pr esence or absence, for instance) outweighs the need for detailed quantification, the beads offer a r a pid concentr ation method, whic h can be automated if the sample volume is lo w ered to 10 ml, enabling high throughput testing (Karthik e yan et al. 2021 , Brighton et al. 2024 ), ho w e v er, small sample volumes may pr e v ent the detection of low abundant viruses.For cases where quantification is important, our results show that PEG precipitation applied on high-volume samples is better able to detect less abundant viruses in RT-qPCR, and facilitates the detection of a greater range of variants in SARS-CoV-2 variant sequencing.

Figure 2 .
Figure2.Intr a-labor atory trials with four users (1-4) processing 45 ml samples with the NMP method and 150 ml and 37.5 ml samples with the PEG method (PEG-150 and PEG-37.5, r espectiv el y) in triplicates for the detection of (A) SARS-CoV-2, (B) crAssphage, (C) Norovirus GI (NoVGI) and (D) Norovirus GII (NoVGII).The boxes show the middle 50% of the data set with the horizontal line r epr esenting the median v alue.Err or bars r epr esent 95% confidence intervals.

Figure 3 .
Figure 3. Vir al concentr ations for crAsspha ge, enter oviruses (EV), influenza A virus (FluA), SARS-CoV-2, norovirus GI and GII (NoV GI, NoV GII), r espir atory syncytial virus (RSV) and Phi6 phage process control virus in w astew ater samples using the NMP (green bars) and the PEG-37.5 (blue bars) and PEG-150 (purple bars) methods for viral concentration.The boxes show the middle 50% of the data set with the horizontal line r epr esenting the median v alue.Err or bars r epr esent 95% confidence interv als.