Eight avian influenza isolates of varying virulence and subtypes ( Table 1 ) from the Southeast Poultry Research Laboratory (SEPRL) were used to validate the performance of the updated protocol for influenza A whole genome sequencing. Viruses were propagated in 9–11-day-old specific-pathogen-free (SPF) embryonated chicken eggs. The harvested allantoic fluids were subsequently used in this study. Background information on the egg-grown isolates is summarized in Table 1 .

Total RNA was extracted from infectious allantoic fluids using the MagMAX™-96 AI/ND Viral RNA Isolation Kit (Applied Biosystems, USA) following the manufacturer’s instructions. RNA quality and concentrations were assessed using the EzDrop 1000C spectrophotometer (Blue-Ray Biotech, Taiwan). The presence of influenza RNA was confirmed using the NVSL avian influenza matrix gene RT-qPCR assay, as previously described (Spackman et al., 2002; Goraichuk et al., 2024a). Extracted viral RNA were then used for the comparison of different RT-PCR kits, primer sets, and purification methods.

To increase the minimum read depth coverage of polymerase segments, we compared the performance of three different RT-PCR kits for the simultaneous amplification of influenza A genome segments. Multisegment RT-PCR amplification was performed according to the ONT Ligation Sequencing Influenza A Whole Genome V14 protocol (Oxford_Nanopore_Technologies, 2024) using the recommended RT-PCR kit (SuperScript™ III One-Step RT-PCR System with Platinum™ Taq DNA Polymerase, Invitrogen, USA) and compared to two other RT-PCR kits with alternative RT and DNA polymerases (SuperScript™ IV One-Step RT-PCR System, Invitrogen, USA and LunaScript^® Multiplex One-Step RT-PCR Kit, New England Biolabs, USA), referred to as SSIII, SSIV, and LS, respectively. The key specifications of the three RT-PCR kits are summarized in Table 2 . Thermocycling conditions and reaction volumes for the ONT-recommended SSIII RT-PCR were performed as described in the ONT Ligation Sequencing Influenza A Whole Genome V14 protocol (Oxford_Nanopore_Technologies, 2024). For the alternative SSIV and LS RT-PCR kits, conditions were adjusted according to the manufacturers’ recommendations and the detailed optimized protocols have been deposited at protocol.io: dx.doi.org/10.17504/protocols.io.bp2l62r15gqe/v1 (Goraichuk et al., 2024b).

Following the RT-PCR kit comparison, the performance of two primer sets was evaluated for the simultaneous amplification of influenza A genome segments to determine which provided higher minimum read depth coverage in polymerase segments. All eight gene segments were amplified using the ONT-recommended Tuni primer set (Zhou et al., 2009) and the alternative Opti primer set (Mena et al., 2016; Leyson et al., 2019) with both the ONT-recommended SSIII and the alternative SSIV RT-PCR kits. Both Tuni and Opti primer sets incorporate influenza Uni 12 and Uni 13 conserved termini at the end of all 8 genomic segments, along with a 10 nt tail at the 5’ end to enhance PCR amplification. The primary difference between the primer sets lies in nucleotide compositions of the 5’ tails, which necessitated adjustments in the annealing temperatures. The sequences of the primers, master mix compositions, and thermocycling conditions are summarized in Supplementary Table 1 . After thermocycling, 5 μL of the product was visualized on a 1.5% agarose gel to verify the amplification of all genomic segments. Additionally, the concentration and purity of the amplicons were measured using the EzDrop 1000C spectrophotometer (Blue-Ray Biotech, Taiwan), Qubit 1X dsDNA High Sensitivity Kit on a Qubit 4 fluorometer (Invitrogen, USA), and High Sensitivity D5000 ScreenTape on a 4150 TapeStation (Agilent Technologies, USA).

To evaluate the effectiveness of different amplicon purification kits in reducing short untargeted reads, we compared two magnetic bead-based and two column-based purification kits. After amplification using the SSIV RT-PCR kit and the Opti primer set, amplicons were purified according to the Nanopore protocol using the bead-based Kit 1 (Agencourt AMPure XP beads, Beckman Coulter, USA) at a 1:1 bead:sample ratio. This was then compared to three alternative purification kits: another magnetic bead-based Kit 2 (ChargeSwitch PCR Clean-Up Kit, Invitrogen, USA) and two column-based kits (Kit 3 – PureLink PCR Purification Kit (Invitrogen, Lithuania) and Kit 4 – Select-a-Size DNA Clean & Concentrator (Zymo Research, USA)) following the manufacturer’s protocols. The elution volumes differed between kits as follows: Kit 1 – 15 µl, Kit 2 - 25 µl, Kit 3 – 50 µl, and Kit 4 – 15 µl.

To improve time efficiency when processing numerous samples, we compared the performance of two bead-based purification methods using both manual and automated processes with the KingFisher Purification System 5400000 (Thermo Scientific, USA). For the automated process, the elution volume was increased to 30 µl according to the manufacturers’ recommendations. Following purification, the concentration and purity of the amplicons were measured using the EzDrop 1000C spectrophotometer (Blue-Ray Biotech, Taiwan), Qubit 1X dsDNA High Sensitivity Kit on a Qubit 4 fluorometer (Invitrogen, USA), and High Sensitivity D5000 ScreenTape on a 4150 TapeStation (Agilent Technologies, USA).

To obtain the reference genomes of the viruses used in this study, we conducted two Illumina sequencing runs using sequence-independent, single-primer amplification (SISPA) library preparation methods (Chrzastek et al., 2017), with and without a previously published pretreatment to remove host and bacterial rRNAs (Parris et al., 2022; Bakre et al., 2023; Goraichuk et al., 2024a). Illumina libraries were prepared using the Illumina DNA Prep (Illumina, USA) according to the manufacturer’s recommendations. After quantification using the Qubit 1X dsDNA High Sensitivity Assay Kit (Invitrogen, USA) and High Sensitivity D5000 Screen Tape (Agilent Technologies, USA), the libraries were pooled (4 nM, 10 µl each), spiked with a control library (5% PhiX library v3), diluted to 12 pM final concentration and sequenced (paired-end; 2x300 bp) using the 600-cycle MiSeq Reagent Kit v3 (Illumina, USA) on an Illumina MiSeq instrument.

For the comparison of different RT-PCR kits, primer sets, purification kits, and purification automation methods, four Nanopore sequencing libraries were prepared using the Native Barcoding Kit 24 V14 (SQK-NBD114.24, Oxford Nanopore Technologies, England). A total of 24 samples per comparison were pooled together after barcoding, and the final library was quantified using the High Sensitivity D5000 Screen Tape on a 4150 TapeStation (Agilent Technologies, USA). We then sequenced 20 fmol of the prepared library for initial RT-PCR comparison, as recommended by the ONT protocol. Subsequently, we increased the loading amount up to 80 fmol in the following runs to achieve higher flow cell pore occupancy and longevity of flow cell ( Supplementary Table 2 ). All comparisons were performed on separate R10.4.1 MinION flow cells (FLO-MIN114, Oxford Nanopore Technologies, England) using the Mk1C sequencer with the MinKNOW 23.04.8 software. Sequencing was run for ~ 24 hours.

The Illumina raw sequencing data was processed within the Galaxy platform. The raw reads from samples prepared using SISPA, both with and without rRNA depletion pretreatment, were merged to enhance the yield of viral reads for complete genome coverage. The forward and reverse raw sequence reads were joined and their quality was assessed using FastQC v0.63 (Andrews, 2023). Low-quality bases were trimmed and short reads were filtered by Fastp 0.32.2 (Chen et al., 2018). Host reads (Gallus gallus and Meleagris gallopavo) were eliminated using the Burrows-Wheeler Alignment Tool (BWA-MEM) (Li and Durbin, 2009), and the output was sorted using Samtools merge 1.15.1 tool (Danecek et al., 2021). Digital normalization via median k-mer abundance was carried out using the BBTools: BBNorm (Bushnell et al., 2017). The remaining unmapped reads were subjected to de novo assembly using the MIRA Assembler v3 (Cock et al., 2013) to obtain an intermediate genome scaffold. The consensus sequence was then re-called by mapping trimmed and filtered paired collection to the genome scaffold using the BWA-MEM (Li and Durbin, 2009). PCR duplicates were removed using RmDup 2.0.1 (Li et al., 2009), and then final consensus sequences were generated using the bam2consensus tool (Volkening, 2023).

The Nanopore raw Pod5 files were basecalled with a high-accuracy algorithm to generate FastQ files, which were then demultiplexed and trimmed using Dorado 7.1.4 within the MinKNOW 23.07.12 (bionic) software on a MinION Mk1C instrument. Reads with a minimum quality of 9 were considered for further analysis. For the RT-PCR kits and primer sets comparisons, short reads below 200 bp were removed during sequencing run. Filtered MinKNOW-generated FastQ files containing 4,000 “pass” reads per file were concatenated into a single consolidated file for each barcoded sample. Further analysis of Nanopore reads was performed on the Galaxy platform. The influenza genome was assembled by aligning filtered reads with reference genomes obtained from Illumina sequencing using minimap2 (Li, 2018) and verified in Geneious Prime 2023.0.1. The coverage of the influenza virus genome was obtained using SAMtools depth (Danecek et al., 2021).

GraphPad Prism 10.2.3 (Sović et al., 2016) was used for data visualization and statistical analysis. A one-way ANOVA followed by Tukey’s multiple comparisons test was utilized to compare the relative difference in the total number of sequenced reads, the mean number of reads, and minimum read depth coverage among different RT-PCR kits, primer sets, and purification kits for the eight viruses sequenced. The p-value ≤ 0.05 was considered statistically significant.

Eight selected influenza A samples were confirmed positive by the NVSL avian influenza matrix gene RT-qPCR assay, with Ct values ranging from 11.6 to 17.6 ( Table 3 ). Two Illumina MiSeq runs generated 1,022,630 to 1,807,144 total raw paired-end reads per sample. De novo assembly resulted in over 99% genome breadth coverage (complete coding genome coverage) of the avian influenza virus in all sequenced samples. The obtained genome sequences were deposited in GenBank under accession numbers PQ060363-PQ060365, PQ064115-PQ064122, PQ064136-PQ064143, PQ064247-PQ064254, PQ064267-PQ064274, PQ064551-PQ064558, PQ106540-PQ106540, EU735794, EU735796, EU735799-EU735801, MH342039 ( Table 1 ) and were used as reference genomes in the optimization of Nanopore sequencing.

In our efforts to optimize the RT-PCR conditions for more efficient amplification of the polymerase genes, we focused on refining the amplification protocol by comparing different RT-PCR kits. To achieve this, we selected two alternative one-step RT-PCR kits: SSIV with the Platinum SuperFi DNA Polymerase and LS with the Q5 Hot Start High-Fidelity, which are more inhibitor-resistant and have a lower error rate compared to the SSIII kit with the Platinum Taq Polymerase recommended in Nanopore’s Ligation Sequencing Influenza Whole Genome Protocol. We further refined the RT-PCR thermocycling conditions for the alternative RT-PCR kits, as they require higher hot-start activation, higher annealing temperature, and shorter annealing time according to the manufacturer’s recommendations. Different RT times, annealing temperature and times were tested (data not shown). The optimal RT-PCR conditions for both alternative kits are provided at dx.doi.org/10.17504/protocols.io.bp2l62r15gqe/v1 (Goraichuk et al., 2024b).

There were no significant differences between the compared RT-PCR kits in the average total number of influenza reads ( Figure 1A ), mean read number ( Figure 1B ), and minimum depth ( Figure 1C ) of reads mapped across the complete influenza genome. However, when examining the polymerase segments (PB2, PB1, and PA) separately, which typically exhibit the lowest minimum read depth, the alternative LS RT-PCR kit provided a significantly higher total and mean number of influenza reads ( Figures 1D, E ). Conversely, the minimum read depth in the polymerase segments was significantly higher (p < 0.05) for amplicons generated with the SSIV RT-PCR kit ( Figure 1F ).

Overall, both alternative RT-PCR kits outperformed the ONT-recommended SSIII kit. The LS kit produced a higher average number of sequenced reads (p < 0.01) and a higher mean read depth (p < 0.0001). However, it exhibited slightly lower minimum read depth coverage in the polymerase segments compared to the SSIV kit, though this difference was not statistically significant. Based on these findings, we selected the SSIV kit for further evaluation, as it provided the highest minimum read depth in the polymerase segments ( Figure 1F ).

Next, we evaluated the performance of two primer sets: the ONT-recommended Tuni set and the alternative Opti set, validating them in conjunction with the ONT-recommended SSIII RT-PCR kit and the alternative SSIV kit, which had previously demonstrated superior results in minimum read depth coverage for polymerase segments. After testing various conditions (data not shown), we determined that an annealing temperature of 66°C for Tuni primers and 67°C for Opti primers was optimal for the alternative SSIV kit ( Supplementary Table 1 ).

Although non-significant, the Opti primer set yielded the highest average total and mean number of mapped reads across the complete genome, regardless of the RT-PCR kit used ( Figures 2A, B ). The average minimum read depth across all segments was higher in samples prepared with an alternative SSIV RT-PCR kit, regardless of the primer set used ( Figure 2C ). In the polymerase segments (PB2, PB1, and PA), there was no significant difference in the total and mean read number of reads between RT-PCR kits ( Figure 2D ). However, when comparing primer sets, the alternative Opti set was superior, providing a statistically significant increase in conjunction with both SSIII and SSIV (p < 0.05 and p < 0.01, respectively) in the average total number of sequenced reads. A similar impact was observed in the average mean read depth across polymerase segments ( Figure 2E ) with the alternative Opti primer set compared to the ONT-recommended Tuni primer set in both SSIII and SSIV RT-PCR kits, although not statically significant. The average minimum read depth of polymerase segments was significantly higher in samples prepared with the alternative Opti primer set in conjunction with the alternative SSIV RT-PCR kit compared to all other combinations ( Figure 2F ). This combination was selected for further evaluation of purification kits.

Notably, the SSIV RT-PCR kit not only increased the minimum read depth of the polymerase segments but also substantially improved the minimum read depth of the HA and NA segments, which are critical for influenza subtyping ( Supplementary Figure 1 ).

Next, we focused on evaluating different purification methods to remove short reads that could originate from defective interfering particles, potentially causing known U-shaped read coverage of polymerase segments. For this, we compared two magnet bead-based kits (Kit 1 and Kit 2) and two column-based kits (Kit 3 and Kit 4). The elution volume for different purification kits varied due to the manufacturer’s recommendations. However, after normalizing the obtained quantities, we found that Kit 1 provided the highest quantity of purified amplicons, followed by Kit 3 and Kit 2, while Kit 4 yielded the lowest quantity – averaging almost 5.7 times lower amount compared to unpurified samples ( Table 4 ). Regarding quality, all purification kits performed well demonstrating A260/A280 absorbance ratio values of 1.8-2.0, which indicates a pure DNA sample. However, Kit 2 and Kit 4 were superior, with an average A260/A280 of 1.8, indicating optimal purity.

The average length distribution measured on the Tape Station indicated that all purification kits effectively eliminated smaller fragments, subsequently increasing the average length of the purified amplicons. Notably, Kit 4 demonstrated a substantial increase in average length compared to the other kits ( Table 4 ). This improvement was also evident in the electropherogram of amplicons distribution before library preparation ( Supplementary Figure 2 ) and further confirmed by the length distribution of sequenced reads ( Figure 3 ).

Purification with Kits 1, 2, and 3 demonstrated that the majority of reads were still below 700 bp, which are considered non-targeted, as the shortest influenza segment is nearly 900 bp. In contrast, purification with Kit 4 effectively removed most of these untargeted reads, resulting in a shift in the read distribution towards longer influenza reads. Despite Kit 4 yielding the least number of reads, these reads had significantly higher median and mean read lengths, as well as a higher N50 value, indicating a successful reduction of shorter reads ( Table 5 ).

The average number of sequenced reads was significantly lower (p < 0.05) after purification with Kit 4 compared to the ONT-recommended and commonly used Kit 1 ( Figure 4A ). However, the average total number of mapped reads, as well as the average mean and minimum read depth across the complete genome, did not show significant fluctuations between purification kits ( Figures 4A–C ). The polymerase segments, however, were more significantly impacted by the purification kits. Specifically, Kit 4 resulted in a significantly lower number of average sequenced reads ( Figure 4D ), while providing significantly higher minimum read depth coverage compared to all other purification kits ( Figure 4F ). The average mean read depth did not show notable variations between purification kits ( Figure 1E ).

The decrease in the total number of sequenced reads with Kit 4 was due to the removal of abundant short reads. This reduction in short reads allowed for a higher proportion of long reads to be sequenced, thereby increasing the minimum read depth of polymerase segments. The shift in read distribution is clear in the influenza genome read coverage plot ( Supplementary Figure 3 ), demonstrating a more uniform coverage across the polymerase segments after purification with Kit 4, contrasting with the results from the other purification kits where the U-shaped distribution persisted. Additionally, Kit 4 offers significant time savings with 2-minute centrifugation and 1-minute incubation, compared to the recommended ONT kit’s 22-minute incubation. Automation with alternative Kit 2 was comparable to the manual purification.

To improve time efficiency for processing a high number of samples, we compared the performance of two bead-based kits (Kit 1 and Kit 2) in manual and automated purification. Our findings demonstrated that the total number of viral reads sequenced across the complete genome decreased with automated purification compared to manual purification ( Figure 5A ). This decrease was statistically significant (p < 0.001) for purification using Kit 1, but not significant for Kit 2. Similarly, the decrease in mean read depth coverage ( Figure 5B ) and minimum read depth coverage ( Figure 5C ) across the genome were significant (p < 0.001 and p < 0.01, respectively) for Kit 1 but not significant for Kit 2. Overall, manual purification with Kit 1 provided superior results.

Interestingly, for the polymerase segments, we observed a similar decrease in total, mean, and minimum read coverage with automation using Kit 1 ( Figures 5D–F ), while there was an increase in reads after automation with Kit 2, although this increase was not statistically significant. Thus, while automation severely affected the performance of the ONT-recommended Kit 1, automation with Kit 2 not only provided comparable results to its manual use but also matched the performance of the manual use of Kit 1. This indicates that Kit 2 is more suitable for automated purification when aiming to maintain high-quality sequencing results while processing a large number of samples efficiently.