Evaluation of the influence of RNA extraction methods on the detection of citrus viruses and viroids Authors: Madelein Dippenaar, Hans Jacob Maree, Rachelle Bester Published: 2025 ABSTRACT BackgroundCertification and disease management practices are reliant on accurate pathogen detection assays. Effective nucleic acid extraction is the cornerstone of various molecular detection techniques, and many virus and viroid RNA extraction protocols have been developed, each with accompanying advantages and limitations. The extraction method influences the type, quantity and quality of RNA and, therefore, the overall sensitivity of the assay.MethodsPlants from two citrus types were infected with multiple viruses and viroids. Ten plants from each were selected. RNA was extracted from bark and leaf samples using four protocols: CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol) and SI (SDS/isopropanol). Extracts were subjected to RT-qPCR for pathogen quantitation. Results were confirmed with Illumina high-throughput sequencing (HTS). Leaf material from a citrus type was resampled, RNA extracted with CE, SI and CP (CTAB/isopropanol) and pathogen quantities determined with RT-qPCR. An additional five plants were also selected and subjected to RNA extraction with CP, SI and a modified CE. RT-qPCRs and HTS were applied to all of these extracts. Protocols were evaluated based on pathogen quantities and various bioinformatic approaches. Reference genes and CTV genotypes were also assessed with HTS.ResultsThe extraction protocol had a remarkable influence on RT-qPCR and HTS pathogen detection ability. The CL method performed best for CTV while SI proved more sensitive for the other viruses. Viroid detection had increased variability with CE extracts containing more viroid RNA although CL extracts showed comparable amounts of CDVd. The CE protocol indicated potential debilitating effects on virus detection. The additional sampled plants subjected to the modified CE protocol, showed increased virus and decreased viroid amounts. The extraction protocol did, however, not significantly influence CTV genotype composition and the investigation of reference genes provided insight into appropriate genes to be used as internal controls for cross sample comparisons.ConclusionsDespite the differential pathogen extraction amongst protocols, all target pathogens could be detected with appropriate RT-qPCR replicates and combined bioinformatic approaches. This study emphasises that although multiple protocols can be effective, further selection and optimisation could enhance the performance. Present research serves as a foundation for the refinement of diagnostic pipelines in citrus pathology.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12985-025-03013-0. Background Plant pathogen detection is imperative for effective disease management and prevention. Agricultural industries are globally threatened by an array of pathogens. Surveillance, certification and quarantine programs are essential to promote the dissemination of healthy and clean propagation material to ensure the sustainable production of commercially important crops [1–5]. For various pathogens, especially pathogens like viruses and viroids, molecular techniques are invaluable to aid rapid and accurate detection. Nucleic acid amplification techniques, most notably PCR and RT-PCR, are extensively validated for plant virus diagnostics and often regarded as the “golden standard” for reliable accuracy and sensitivity. Recently, the use of high-throughput sequencing (HTS) has gained traction as an alternative to conventional detection methods for routine diagnostics [1, 2, 5–9]. The ability to detect novel pathogens, without any prior knowledge or references and the constant improvement and development offering scalable solutions, has led to increased interest worldwide in this technology for application in various plant disease prevention and management strategies. PCR methods and HTS are, however, influenced by the efficient extraction of nucleic acid from the plant material and can directly affect the outcome of pathogen identification. Extraction protocols have different limitations and criteria, and as a result, the type, quality, and quantity of nucleic acid extracted varies significantly [1, 10–19]. RNA extraction is often employed for pathogen identification, but given the diversity of viruses and viroids, the extraction method can have adverse effects when applying the same protocol for multiple pathogen detection. Plant viruses alone are extensive in characteristics with a large size range and multiple genome compositions, but in conjunction with viroids, the smallest known plant pathogen of only a few hundred nucleotides composed in a circular genome [20, 21], it is understandable that protocols may have different detection capabilities. Extraction protocols are often optimised for viroid RNA because their low concentration and uneven distribution in plants can prevent repeatable and consistent detection of these pathogens [1, 2]. Historically, the method of RNA precipitation has been implicated as a limiting factor for small RNA extraction. Specifically, the diminishment of viroid RNA extraction caused by lithium chloride precipitation at certain concentrations compared to other protocols has been investigated and corroborated in multiple studies [1, 2, 11, 12, 22]. Conversely, ethanol precipitation has been shown to enhance small RNA extraction. Precipitation with alcohol generally yields total nucleic acid and despite specially optimised RNA extraction protocols, genomic DNA contamination is usually unavoidable [23]. This can be mitigated by the use of RNase-free DNase treatment, but it is time-consuming and has an increased risk of RNA degradation [23, 24]. Various protocols indicated to have adequate viroid RNA levels in extracts, have been matrix-based approaches that employ columns for isolation and purification [1, 12, 14, 25, 26]. Matrix-based protocols are often linked to commercially available kits [26, 27] but one limiting factor often associated with these kits is the small amount of input material generally used. This might be beneficial in some cases, but when representative sampling is crucial, as is the case with unevenly distributed pathogens, scalable solutions that can accommodate larger input material amounts are often required [1]. The severity of differences between protocols needs to be assessed appropriately within specific contexts in order to determine the significance thereof, especially when considering the implementation in routine practices. It is, therefore, important to consider the pathogen scope. Citrus is a natural host for a wide range of pathogens including large viruses like citrus tristeza virus (CTV, 19.3 kb) and viroids (291–475 nt). It is, therefore, important to continuously evaluate and validate existing pathogen management strategies, while also exploring alternative approaches, to ensure that the most suitable methods for routine citrus pathogen detection are applied. The identification of a practical protocol well suited towards integration into existing but also possible new strategies, that can enable the reliable detection of all relevant RNA viruses and viroids from citrus material is key for the production of healthy plant material and the sustainable production of citrus. Various citrus virus and viroid RNA extraction protocols, each with accompanying advantages and limitations, have been developed and evaluated [1, 2, 12, 14, 26], but further investigation with larger sampling volumes, across citrus and tissue types are required to establish a pragmatic extraction protocol equally suited for virus and viroid RNA detection. Additionally, various of these protocols use commercially available kits or buffers, and the previously mentioned limitations thereof, regarding sample input and cost, necessitate the development of reliable and consistent internal laboratory solutions. Building on previous studies of citrus pathogen HTS detection, that highlighted the variation and potential for improvement in the simultaneous extraction of viruses and viroids, this study set out to investigate two citrus species infected with a range of viruses and viroids to evaluate different RNA extraction protocols. Virus and viroid quantities were determined with RT-qPCR assays in bark and leaf extracts and results were confirmed with a selected group using Illumina HTS. This allowed not only the comparison of different protocols across citrus types but also enabled the direct comparison of RT-qPCR and HTS for pathogen detection and extraction protocol suitability. Methods Plant material, screening and selection Thirty seedlings of two citrus types each, Citrus sinensis (L.) Osbeck cv. ‘Madam Vinous’ sweet orange and C. aurantiifolia (Christm.) Swingle cv. ‘Mexican Lime,’ were graft inoculated with a well-characterised source plant that contains a range of pathogens including multiple viruses and viroids [1, 2] (Fig. 1). All plants were subjected to screening six months post inoculation for two viruses and three viroids using RT-PCR assays. The viruses include Closterovirus tristezae, commonly referred to as citrus tristeza virus (CTV), and Coguvirus eburi, referred to as citrus virus A (CiVA) [28]. The viroids include Apscaviroid nanocitri, known as citrus dwarfing viroid (CDVd), Pospiviroid exocortiscitri, or citrus exocortis viroid (CEVd) and Hostuviroid impedihumuli, commonly known as hop stunt viroid (HSVd) [28]. These pathogens were selected to cover a range of graft transmissible pathogens including a positive and a negative sense RNA virus as well as three viroids, each from a different genus. Screening of these pathogens (Additional file 1) as well as genotyping for eight CTV genotypes (Additional file 2) were completed using RT-PCR assays and methodologies as described by Bester et al. 2021 [1]. Results obtained from screening and genotyping were used to inform the selection of ten plants from each citrus type. Both bark (Madam Vinous: 0.94 g ± 0.11; Mexican Lime: 0.87 g ± 0.14) and leaf material (Madam Vinous: 0.41 g ± 0.11; Mexican Lime: 0.28 g ± 0.13) from each selected plant were sampled ten months post inoculation for RNA extractions and this collection of samples was designated as Group A (Fig. 1). Following the acquisition and analysis of Group A’s results, two additional samplings of plant material occurred seventeen months post inoculation. The first, Group B, consisted of leaf material from the ten originally selected Madam Vinous (MV) plants (0.44 g ± 0.07). Then, for Group C, five new MV plants were selected, based on the screening and genotyping results previously obtained, and leaf material sampled (0.39 g ± 0.09). Fig. 1Graphical overview of methods and general workflow. All inoculated seedlings were initially screened, and CTV genotyped after which 3 rounds of selection took place as indicated by Group A – C. After results obtained from Group A, the MV plants were resampled, and a new extraction protocol was included along with two previously investigated protocols (Group B). Initial results from this group led to the selection of 5 new plants (Group C). These plants were subjected to the same extraction protocols as Group B, but the CE protocol was modified in this instance. All RNA extracts were subjected to pathogen RT-qPCR assays. Selected samples from Group A and all samples from Group C were sent for HTS. MV – Madam Vinous; ML – Mexican Lime; CL – CTAB/lithium chloride protocol; CE – CTAB/ethanol protocol; TL – acid phenol buffer/lithium chloride; SI – SDS/isopropanol; CP – CTAB/isopropanol Graphical overview of methods and general workflow. All inoculated seedlings were initially screened, and CTV genotyped after which 3 rounds of selection took place as indicated by Group A – C. After results obtained from Group A, the MV plants were resampled, and a new extraction protocol was included along with two previously investigated protocols (Group B). Initial results from this group led to the selection of 5 new plants (Group C). These plants were subjected to the same extraction protocols as Group B, but the CE protocol was modified in this instance. All RNA extracts were subjected to pathogen RT-qPCR assays. Selected samples from Group A and all samples from Group C were sent for HTS. MV – Madam Vinous; ML – Mexican Lime; CL – CTAB/lithium chloride protocol; CE – CTAB/ethanol protocol; TL – acid phenol buffer/lithium chloride; SI – SDS/isopropanol; CP – CTAB/isopropanol RNA extraction protocols Initially, four RNA extraction protocols were selected (Fig. 2) including a modified cetyltrimethylammonium bromide (CTAB) protocol [29, 30] with lithium chloride (LiCl) precipitation (abbreviated as CL). The second method (abbreviated as CE) followed the same protocol but utilised 2.5 times absolute ethanol and 0.1 times 3 M sodium acetate incubated at −20 °C for an hour to precipitate total nucleic acid [31]. The third method, described by Cook et al. (2016) [32], used an acidic phenol extraction buffer and precipitated total RNA with both isopropyl alcohol and LiCl (abbreviated as TL). Lastly, a total nucleic acid extraction protocol with a sodium dodecyl sulphate (SDS) buffer and isopropanol precipitation (abbreviated as SI) was adapted from Arruabarrena et al. (2016) and Oñate-Sánchez and Vicente-Carbajosa (2008) [33, 34]. Appropriate modifications were made to accommodate a larger amount of input plant material but the content and concentration of all buffers and solutions remained the same. Briefly, plant material was homogenised in 5 ml cell lysis solution (pH 4) and incubated at room temperature for 5 min. The plant liquid (2 ml) was transferred to clean microcentrifuge tubes and centrifuged (17 136 rcf) at room temperature for 1 min to pellet debris. The supernatant (1.2 ml) was transferred to a new microcentrifuge tube and 400 µl of protein-DNA-precipitation solution was added. These tubes were then incubated on ice for 10 min and centrifuged (17 136 rcf) for an hour at 4 °C. The supernatant was then transferred to new microcentrifuge tubes (1 ml) and total nucleic acid was precipitated using 1x isopropanol. Each sample from Group A was subjected to each of these protocols (Figs. 1 and 2). The resampled MV plants (Group B) underwent RNA extraction with three protocols including both CE and SI mentioned previously and a new protocol (Figs. 1 and 2). Similar to CE, the new protocol (abbreviated as CP) followed the CL protocol but with a modification in nucleic acid precipitation. In this instance 1x isopropanol was added to precipitate nucleic acids. Leaf material from five additional plants (Group C) was also subjected to these three protocols (CE, SI and CP) but with a slight adjustment made in the CE protocol. In Group A, CE indicated severe bias towards small RNAs when plant material homogenised in extraction buffer was exposed to cold temperatures in the presence of chloroform:isoamyl alcohol (C:I) for an extended period of time (more than 15 min) during purification. To increase the amount of larger RNA molecules like ribosomal RNA (rRNA) extracted with this protocol, modifications in protocol execution and streamlining of sample processing were made. This included processing less samples at a time (five instead of ten) to reduce sample exposure to conditions prior to transfer and/or centrifugation protocol steps. This especially applied to the first C:I purification step where the homogenised plant liquid (after water bath incubation) was transferred to cold C:I containing microcentrifuge tubes (on ice). The transfer time, in which the finished tubes waited on ice, was significantly reduced from over 15 min to under five minutes through reduced sample processing and the addition of 5 ml CTAB buffer instead of 4 ml, as used in Group A and C, to ease transfer. Therefore, the extraction protocols selected included total RNA extraction protocols CL and TL and three total nucleic acid extraction protocols SI, CE and CP (Fig. 2). The TL protocol is currently integrated into the improvement program of commercial citrus in South Africa where TL is the standard protocol used for diagnostics with RT-PCR. The CL protocol has been previously evaluated for potential inclusion in the improvement program [1, 2]. Total nucleic acid protocols were investigated for their proficiency in extracting small RNAs since both CL and TL precipitate RNA using LiCl which is not as efficient for small RNA precipitation [1, 11, 22]. Ethanol precipitation has also previously been shown to be more effective at citrus viroid RNA extraction, when used in conjunction with a CTAB extraction protocol with acidic phenol phase separation, compared to the CL protocol as described here [2]. The SI protocol was selected due to advantages such as short extraction time, relatively affordable reagents and the use of non-hazardous chemicals [33, 34] unlike chloroform, beta-mercaptoethanol (βME), phenol and guanidine thiocyanate used in the other protocols. Since this protocol applied isopropanol for precipitation, the CP protocol was also investigated to enable a buffer comparison between CTAB and the SDS buffer of SI. Fig. 2RNA extraction methods summary. Key considerations highlighting the differences between protocols include the type of extraction, the buffer and the method of precipitation. The Group(s) in which the protocol was used is also indicated. CTAB - cetyltrimethylammonium bromide; LiCl – lithium chloride; GITC - guanidinium thiocyanate; CL – CTAB/lithium chloride protocol; CE – CTAB/ethanol protocol; TL – acid phenol buffer/lithium chloride; SI – SDS/isopropanol; CP – CTAB/isopropanol. The modified CE protocol (*CE) as applicable to Group C had the same buffer and precipitation method as the unmodified CE protocol (Group A and B) RNA extraction methods summary. Key considerations highlighting the differences between protocols include the type of extraction, the buffer and the method of precipitation. The Group(s) in which the protocol was used is also indicated. CTAB - cetyltrimethylammonium bromide; LiCl – lithium chloride; GITC - guanidinium thiocyanate; CL – CTAB/lithium chloride protocol; CE – CTAB/ethanol protocol; TL – acid phenol buffer/lithium chloride; SI – SDS/isopropanol; CP – CTAB/isopropanol. The modified CE protocol (*CE) as applicable to Group C had the same buffer and precipitation method as the unmodified CE protocol (Group A and B) Virus and viroid quantitation with RT-qPCR Absolute quantitation using standard curves generated from gBlock Gene Fragments (IDT) were used to determine the precise copy number of each pathogen present in each sample. Pathogens quantitated in the RT-qPCR analyses, CTV, CiVA, CDVd, CEVd and HSVd, were the same as those included in the initial screening. These gBlock Gene Fragments were designed based on the RT-qPCR amplicon sequences as designated by the specific primers (Additional file 4). Additional nucleotides were added to both ends of the gBlock sequences for improved primer binding and amplification (Additional file 3). A 10-fold dilution series with nine dilution points starting at 0.1 ng/µl for each gBlock Gene Fragment was prepared with water and used to construct a standard curve. The exact copy number of each dilution point was calculated with the formula in Fig. 3. For virus and viroid quantitation each extract from Group A, B and C were subjected to two-step RT-qPCRs. cDNA synthesis was carried out as previously described for RT-PCR screening, following the methodology described by Bester et al. 2021 [1]. This was followed by amplification using 1 µl diluted (x25) template cDNA made from 1 µg of RNA. A negative control sample, non-template control, gBlock dilution point four (0.0001 ng/µl) and a diluted (x25) pooled cDNA sample representing all extracts and consequently positive for all pathogens were included in all runs. Real-time amplification of the cDNA dilution series and samples were executed in triplicate using 1x SensiFAST SYBR No-ROX Mix (Bioline Reagents Ltd., London) with 0.4 µM of forward and reverse primer (Additional file 4) in a total reaction volume of 12.5 µl. Reactions were carried out in the QuantStudio 5 real-time PCR system (Applied Biosystems). After completion of the temperature cycling, a melt curve analysis using the system default settings with temperatures ranging from 60 °C to 95 °C proceeded. The Design and Analysis software version 2.6.0 (Applied Biosystems), for the QuantStudio 5 system, was used to calculate the pathogen standard curves metrics as well as virus and viroid copy numbers across samples. All extracts’ copy numbers were determined using the same standard curves. Statistical analyses were conducted with the R statistical computing environment (R Core Team, http://www.R-project.org/) [35] using the non-parametric Kruskal–Wallis rank sum test [36, 37] followed by multiple comparison testing with Dunn’s test if appropriate [38, 39]. All graphs were constructed using the package ggplot2 [40] available from the Comprehensive R Archive Network (CRAN). Fig. 3Copy number formula to calculate the number of molecules present in each dilution point of the gBlock standard curve. X – the amount of amplicon (ng), N – length of dsDNA amplicon, 660 g/mol – the average mass of 1 bp dsDNA, 6.022 × 1023 – Avogadro’s constant, 1 × 109 – conversion factor Copy number formula to calculate the number of molecules present in each dilution point of the gBlock standard curve. X – the amount of amplicon (ng), N – length of dsDNA amplicon, 660 g/mol – the average mass of 1 bp dsDNA, 6.022 × 1023 – Avogadro’s constant, 1 × 109 – conversion factor High-throughput sequencing and data quality assessment Five plants of each citrus type in Group A (Fig. 1), MV and Mexican Lime (ML), were selected based on RT-qPCR results. All four leaf extracts from these ten plants and all extracts from all plants in Group C were sent for HTS. A total of 55 ribo-depleted RNA libraries were constructed with the Illumina TruSeq Stranded Total RNA Sample Preparation kit with Plant Ribo-Zero at Macrogen (South Korea) and paired-end HTS (2 × 150 bp) was performed on an Illumina NovaSeqX instrument (Macrogen, South Korea). Before proceeding with data analysis, quality trimming and adapter sequence removal were carried out using Trimmomatic 0.39 [41] (SLIDINGWINDOW of 3 nts with Q20, MINLEN of 20 nts) [30]. Sequencing data quality was assessed before and after trimming with FastQC 0.12.1 [42] and SeqKit 2.7.0 [43]. De novo assembly Trimmed sequencing data was de novo assembled with SPAdes 3.14 [44] using default parameters. The assembled contigs were identified using BLAST + standalone against a local copy of the NCBI GenBank nucleotide database (downloaded January 2025) using the Blastn algorithm [1, 30]. Blastn hits were categorised into four groups to gain insight into the read composition of assembled datasets. These included hits identified as virus, viroid, unidentified and other which referred to all other types of hits as well as host genome. Read mapping The quality trimmed sequencing data were mapped to reference sequences obtained from Genbank in three separate instances. All mapping strategies followed the same general pipeline wherein the Burrows-Wheeler Alignment Tool (BWA) 0.7.18 was utilised [45], followed by filtering of the mapped reads and calculation of the genome coverage and depth using samtools 1.20 [30, 46]. Firstly, host reference gene mapping was performed using twelve genes previously indicated as suitable C. sinensis reference genes [47] (Additional file 5). CTV genotypes of each sample were then determined using eleven CTV genotypes [48] and more stringent mapped read filtering of 95% identity over 90% of the read [30]. Lastly, reads were mapped to pathogen reference genomes to have an indication of the number of reads associated with each pathogen across different protocols (Additional file 5). Genomes included in this reference mapping analysis corresponded to the pathogen profile of the source plant which included the screened pathogens previously mentioned, as well as Capillovirus mali or citrus tatter leaf virus (CTLV) [28]. The CTV genotyping results informed the selection of CTV references included in pathogen read mapping. All genotypes with a genome coverage above 50% were included as 50–90% coverage indicates the presence of genotype variants not represented in the mapping reference list [30]. In all read mapping approaches, the read counts were normalised by calculating the transcripts per million (TPM) [1]. This enabled the comparison between samples based on the proportion of reads mapped to a specific reference sequence. For pathogen read mapping, both CiVA genomes were used to compare the protocols’ performance with HTS but comparisons between sequencing and RT-qPCR were made using only CiVA-RNA1 because the assay’s primer binding sites were located on this genome. Where appropriate, significance testing was carried out in R [35] with the Kruskal–Wallis rank sum test [36, 37] and graphs and principle component analyses were visualised using the packages ggplot2 [40] and ggbiplot [49]. In addition to pathogen read mapping, the consensus sequences generated from the read mapping results of each sample were extracted using samtools and bcftools 1.19 [50]. These consensus sequences were then aligned to the pathogen reference genomes using MAFFT 7.49 [51] and the sequence identities calculated. Results Sample selection RT-PCR was used to screen all Madam Vinous and Mexican Lime plants for CTV, CiVA, CDVd, CEVd and HSVd as well as to determine the CTV genotypes present (Additional file 6). All Madam Vinous plants except for two were positive for the complete pathogen range. The Mexican Lime plants showed more variability regarding infection status and selection preference was given to plants of adequate sampling size containing CTV and at least two other pathogens. Ten plants (Group A) were then selected from Madam Vinous; MV1, MV2, MV4, MV5, MV8, MV9, MV18, MV20 and MV28; and from Mexican Lime; ML1, ML2, ML7, ML9, ML11, ML15, ML16, ML18, ML26 and ML27. The same ten Madam Vinous plants from Group A were resampled as Group B and an additional five Madam Vinous plants; MV3, MV6, MV11, MV19 and MV30; were also selected as part of Group C. RNA extractions Four RNA extraction protocols were initially evaluated with samples from Group A which included leaf and bark material. Of these protocols evaluated, the CE extracts had the lowest purity ratios (Table 1). Interestingly, the SI and TL extracts had similar RNA yield (µg RNA/mg input plant material) and while the CL extracts had the lowest yield, the purity of extracts were comparable to that of SI and TL extracts. Considering the extraction results of Group B and C, similarities were observed between these groups and Group A with regards to the SI extracts. However, the CE extracts showed noticeable differences between Group B and C where the protocol modification led to an increase in yield and quality as expected. The CP extracts included in Groups B and C had consistent results across groups and performed comparably to the CE extracts when the modification was applied (Group C). Table 1Purity ratios and normalised RNA amount across extraction protocols and tissue types for Groups A-CGroupABCProtocol aTLSICLCECECPSICE*CPSITissue TypeLeafBarkLeafBarkLeafBarkLeafBarkLeafLeafLeafLeafLeafLeaf 260/230 MV bAverage1.61.82.22.02.12.01.11.21.11.82.21.71.92.3ML cAverage1.92.12.21.91.92.21.21.1 260/280 MVAverage2.12.12.01.92.12.11.81.91.92.01.92.02.01.9MLAverage2.12.11.91.92.12.11.91.8 RNA Amount (µg RNA/mg input) MVAverage0.6450.5460.6170.5340.1390.1060.1520.3430.1810.3370.8480.4140.3810.725Standard Deviation0.1320.0740.0970.0590.0650.0260.0690.2120.1270.1380.1080.1480.1420.142MLAverage0.8730.6420.7850.5910.0620.0940.4050.456Standard Deviation0.0810.1420.1400.1380.0560.0700.0810.169a Protocols include CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE* protocol refers to the modified CE protocolb Madam Vinousc Mexican Lime Table 1Purity ratios and normalised RNA amount across extraction protocols and tissue types for Groups A-CGroupABCProtocol aTLSICLCECECPSICE*CPSITissue TypeLeafBarkLeafBarkLeafBarkLeafBarkLeafLeafLeafLeafLeafLeaf 260/230 MV bAverage1.61.82.22.02.12.01.11.21.11.82.21.71.92.3ML cAverage1.92.12.21.91.92.21.21.1 260/280 MVAverage2.12.12.01.92.12.11.81.91.92.01.92.02.01.9MLAverage2.12.11.91.92.12.11.91.8 RNA Amount (µg RNA/mg input) MVAverage0.6450.5460.6170.5340.1390.1060.1520.3430.1810.3370.8480.4140.3810.725Standard Deviation0.1320.0740.0970.0590.0650.0260.0690.2120.1270.1380.1080.1480.1420.142MLAverage0.8730.6420.7850.5910.0620.0940.4050.456Standard Deviation0.0810.1420.1400.1380.0560.0700.0810.169a Protocols include CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE* protocol refers to the modified CE protocolb Madam Vinousc Mexican Lime Purity ratios and normalised RNA amount across extraction protocols and tissue types for Groups A-C a Protocols include CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE* protocol refers to the modified CE protocol b Madam Vinous c Mexican Lime Pathogen quantitation The copy number of each pathogen was determined for all extracts using RT-qPCR with absolute quantitation. When considering the four protocols evaluated with Group A (Fig. 1), there were clear distinctions between the protocols’ suitability for differential pathogen detection (Fig. 4). The infection status of plants mostly conformed to what was observed with initial screening with some notable exceptions including no amplification of CiVA in CE extracted MV4 bark material of Group A. The remaining discrepancies were attributed to Mexican Lime samples from Group A where low viroid copy numbers, especially in leaf samples, often led to differential amplification outcomes regardless of extraction protocol used. Group B and C extracts had amplification of all pathogens as indicated with initial screening results. The dominant trend observed for CTV across all data subdivisions, regardless of citrus or tissue type, was the outperformance of CL extracts in comparison to the other three protocols’ extracts (Fig. 4a). In most cases CL extracts had significantly higher CTV copy numbers, while other protocols’ extracts had more comparable quantities. Notably, there was a dissimilarity in the protocol that generated extracts with the lowest CTV quantity between tissue types. While the CE extracts performed the poorest for leaf material, the TL protocol seemed less suited for bark, producing extracts with the lowest copy number in this tissue type. CTV results were starkly contrasted by the results of CiVA (Fig. 4b), the other virus evaluated. The CTAB extracts including CL and CE showed inferior CiVA quantity compared to that of SI and TL extracts. While CE extracts had considerably lower copy numbers than CL, the SI and TL extracts had significantly higher CiVA results compared to the extracts from both CTAB protocols for most of the subdivisions. For the viroids, CDVd (Fig. 4c) showed more variability in observable trends across protocols than the other pathogens. Overall, both CL and CE extracts were implicated in achieving high quantities, although in Madam Vinous the CE extracts had the highest copy numbers compared to Mexican Lime where the CL extracts resulted in the highest copy numbers. SI and TL extracts produced similar results for CDVd. CEVd results were somewhat encumbered by the low concentration of this viroid present in Mexican Lime samples (Fig. 4d). Nonetheless, results from Madam Vinous samples showed that CE extracts had significantly higher CEVd copy numbers compared to extracts from the other extraction methods. The other three methods generated similar results with CL and SI extracts sometimes yielding higher quantities although not statistically significant. HSVd results (Fig. 4e) showed good resemblance to the trends observed with CEVd, however, the CE protocol seemed to have a stronger affinity for HSVd, especially with Madam Vinous samples, as the differences between the yield of CE extracts and other protocols’ extracts were more severe. Regarding the overall viroid protocol evaluation, the CE extracts generated the most desirable results, followed by CL extracts, while SI and TL extracts had the lowest viroid copy number results. Similar to Group A (Fig. 1), virus and viroid quantities were also determined with extracts from Group B and C which included an additional RNA extraction method (Fig. 5). The results obtained relative between the CE and SI protocols from Madam Vinous Group A leaf extracts were confirmed across all pathogens with the re-sampling results of Group B (Fig. 5). SI extracts had significantly higher copy numbers for viruses (Fig. 5a and b) and CE extracts for viroids (Fig. 5c-e). The CP extracts, had results that performed between CE and SI extracts for Group C, never yielding the highest or lowest quantity. Contrary to these results, Group C showed an increase in virus quantity and a decrease in viroid quantity obtained with extracts from the modified CE protocol, therefore, making CE and CP extracts comparable. The SI protocol in the context of Group C extracted significantly higher copy numbers for nearly all pathogens. With regards to all RT-qPCR runs, no contamination, as determined by the melting temperature profiles generated from melt curve analysis, was observed in the negative and non-template controls. Fig. 4Average copy number associated with each protocol across samples in Group A are displayed in a for citrus tristeza virus (CTV), in b for citrus virus A (CiVA), in c for citrus dwarfing viroid (CDVd), in d for citrus exocortis viroid (CEVd) and in e for hop stunt viroid (HSVd). The data is grouped according to different subdivisions including the overall dataset, tissue type (leaves and bark), citrus variety (Madam Vinous, MV, and Mexican Lime, ML) and tissue type within citrus variety. Letters a-c shown on top of the error bars indicate the statistical significance between protocols within a subdivision. Similar letters are indicative of non-significant differences (p > 0.05). Protocols are listed on the right as CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol) and SI (SDS/isopropanol) Average copy number associated with each protocol across samples in Group A are displayed in a for citrus tristeza virus (CTV), in b for citrus virus A (CiVA), in c for citrus dwarfing viroid (CDVd), in d for citrus exocortis viroid (CEVd) and in e for hop stunt viroid (HSVd). The data is grouped according to different subdivisions including the overall dataset, tissue type (leaves and bark), citrus variety (Madam Vinous, MV, and Mexican Lime, ML) and tissue type within citrus variety. Letters a-c shown on top of the error bars indicate the statistical significance between protocols within a subdivision. Similar letters are indicative of non-significant differences (p > 0.05). Protocols are listed on the right as CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol) and SI (SDS/isopropanol) Fig. 5Average copy number associated with each protocol across samples in Group A-C are displayed in a for citrus tristeza virus (CTV), in b for citrus virus A (CiVA), in c for citrus dwarfing viroid (CDVd), in d for citrus exocortis viroid (CEVd) and in e for hop stunt (HSVd). Samples designated as Group A only refer to Madam Vinous leaf sample extracts from Group A so as to be comparable with Group B. Letters a-c shown on top of the error bars indicate the statistical significance between protocols within a group. Similar letters are indicative of non-significant differences (p > 0.05). Protocols are listed on the right as CE (CTAB/ethanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE protocol as applicable to Group C refers to the modified CE protocol Average copy number associated with each protocol across samples in Group A-C are displayed in a for citrus tristeza virus (CTV), in b for citrus virus A (CiVA), in c for citrus dwarfing viroid (CDVd), in d for citrus exocortis viroid (CEVd) and in e for hop stunt (HSVd). Samples designated as Group A only refer to Madam Vinous leaf sample extracts from Group A so as to be comparable with Group B. Letters a-c shown on top of the error bars indicate the statistical significance between protocols within a group. Similar letters are indicative of non-significant differences (p > 0.05). Protocols are listed on the right as CE (CTAB/ethanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE protocol as applicable to Group C refers to the modified CE protocol HTS sample selection Based on the RT-qPCR findings, five leaf samples from each citrus type of Group A were selected for HTS. The Madam Vinous samples selected were MV4, MV5, MV8, MV9 and MV28 and the Mexican Lime samples were ML9, ML15, ML18, ML26 and ML27 (Additional file 6). The average RIN value for all ML extracts across protocols were lower than the MV counterparts, and only the CL extracts had an adequate RIN average (7.43 ± 0.41) (Table 2). Although the MV extracts had better RIN values, CE extracts’ average value was still low compared to other protocols’ extracts (3.67 ± 0.63). When looking at the rRNA ratios, the total nucleic acid extraction protocol extracts, CE and SI, had lower ratios, although it is worth mentioning that SI extracts’ ratios were considerably higher than that of CE extracts. All samples from Group C were also sent for sequencing which included extracts from CP, SI and modified CE protocols. It should, however, be noted that Group C comprised of different MV plants than Group A so some variation due to biological factors was expected. All protocols evaluated from Group C were also total nucleic acid extraction protocols. Nonetheless, both SI and CE extracts had improved RIN values and rRNA ratios compared to previous MV results (Table 2) although the modification of the CE protocol caused a more severe increase than that observed for the SI extracts. The CP protocol performed well producing extracts with the highest average rRNA ratio across all protocols. Table 2HTS sample quality control metrics prior to library preparationGroupSamplesProtocolaRINrRNA ratioAverageSt. DeviationAverageSt. Deviation A Madam VinousTL7.910.361.780.08SI6.440.741.080.22CL8.540.171.250.42CE3.670.630.100.00Mexican LimeTL5.670.941.330.27SI4.301.220.680.26CL7.430.411.020.04CE3.410.650.260.16 C Madam VinousCE*8.220.231.820.04CP7.800.202.060.05SI7.280.291.360.15a Protocols include CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE* protocol refers to the modified CE protocol Table 2HTS sample quality control metrics prior to library preparationGroupSamplesProtocolaRINrRNA ratioAverageSt. DeviationAverageSt. Deviation A Madam VinousTL7.910.361.780.08SI6.440.741.080.22CL8.540.171.250.42CE3.670.630.100.00Mexican LimeTL5.670.941.330.27SI4.301.220.680.26CL7.430.411.020.04CE3.410.650.260.16 C Madam VinousCE*8.220.231.820.04CP7.800.202.060.05SI7.280.291.360.15a Protocols include CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE* protocol refers to the modified CE protocol HTS sample quality control metrics prior to library preparation a Protocols include CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE* protocol refers to the modified CE protocol Sequencing data quality Illumina sequencing generated high quality data with an overall average of 24.8 million paired-end reads corresponding to an average of 3.7 giga bases per sample (Table 3). Average quality scores of reads, irrespective of group, were similar across protocols and consequently 95% of data was retained after quality trimming and adapter removal. CL extracts had the highest data retention (96%) and when comparing Group A and C, the CE extracts had the lowest data retention despite the modifications made in Group C. Table 3HTS data quality summary and data retention (bases) after trimming across protocols for Group A and CGroupProtocol aTotal readsTotal gigabasesAverage read qualityData retention after trimming A CL24,052,8353.63596%CE25,346,8273.83494%TL24,218,5423.73595%SI25,672,1613.93495% C CE*23,255,4283.53593%CP24,942,5553.83594%SI25,632,8093.93694%a Protocols include CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE* protocol refers to the modified CE protocol Table 3HTS data quality summary and data retention (bases) after trimming across protocols for Group A and CGroupProtocol aTotal readsTotal gigabasesAverage read qualityData retention after trimming A CL24,052,8353.63596%CE25,346,8273.83494%TL24,218,5423.73595%SI25,672,1613.93495% C CE*23,255,4283.53593%CP24,942,5553.83594%SI25,632,8093.93694%a Protocols include CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE* protocol refers to the modified CE protocol HTS data quality summary and data retention (bases) after trimming across protocols for Group A and C a Protocols include CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE* protocol refers to the modified CE protocol De novo assembly Sequencing data was de novo assembled using SPAdes followed by Blastn identification of contigs. Assembly of Group A samples produced an average of 133,000 contigs across all protocols, with CL extracts generating the most at an average of 155,000 with a N50 of 891.3 (Table 4). The CE extracts produced the fewest contigs and exhibited the lowest average N50 (410.6), while CL extracts’ average was more than double that of CE. An average of 156,000 contigs were assembled across protocols in Group C (Table 4). Compared to Group A’s CE protocol, the extracts from the modified CE protocol from Group C had a considerable increase in contig assembly as well as N50 (842.6). The same was observed for the SI extracts which also had the highest average N50. Interestingly, although CP extracts had more assembled contigs than CE, the average N50 of CP extracts was the lowest in Group C. For Group A, contig identity composition showed variation between plants within the same extraction group. This variation was more evident for Mexican Lime samples which could have been a consequence of low pathogen concentrations and the differential infection status of plants, while all Madam Vinous plants were positive for all pathogens. Regardless, for both Madam Vinous and Mexican Lime the CE extracts had the lowest virus component (4.3% and 2.1%) (followed by TL extracts for Madam Vinous samples) and the CL extracts had the highest virus component in Madam Vinous samples (10.6%), but the variation in pathogen status of Mexican Lime samples led to similar performances from CL, TL and SI extracts (Table 5). Conversely, for Madam Vinous samples, the viroid component was the highest for CE (0.09%), the lowest for TL (0.006%) and comparable for CL and SI (0.016% and 0.015%) extracts. The Mexican Lime samples, however, had similar viroid components across protocols (0.003%). Less variation was observed in the contig identity composition across biological replicates and protocols from Group C (Table 5). The identity composition of assembled contigs was comparable for all extraction protocols, including the modified CE method, and the virus component was larger than previously seen with Group A with an overall average of 14.5%. Although the viroid component accounted for a very small portion of the assembled data, it is worth noting that while CE and CP extracts had similar results (0.036% and 0.037%), SI extracts had less viroid contigs at 0.014%. This result obtained with the SI protocol was consistent with that observed in Group A, while the modified CE extracts had less viroid contigs in Group C compared to CE extracts in Group A. When considering the detection of the specific pathogen profile, not all protocols were able to consistently detect all pathogens using de novo assembled contigs and Blastn in both Group A and C (Additional file 7). Mexican Lime samples had some contradictions in pathogen identification compared to results obtained from the initial screening (Additional file 6). At least one sample from each extraction group did not have assembled contigs corresponding to all pathogens that screening indicated as present. The results from Madam Vinous samples in both Group A and C were more consistent. In Group A only CE and TL extracts did not result in contigs with similarity to the full pathogen compliment in all samples. In Group C only CP extracts had results mirroring the screening with both the modified CE and SI extracts not resulting in contigs with identity to all of the pathogens in some samples. For both groups, HSVd proved to be the most elusive and even in the Mexican Lime plants, contigs with identity to this viroid was often not assembled even though the initial screening indicated a positive result. Table 4De novo assembled average contig numbers and N50 across protocols for Group A and CGroupProtocol aAverage number of contigsAverage N50 A CL155375.3891.3CE101836.5410.6TL129002.6821.3SI145,718686.4 C CE*136432.8842.6CP171464.0798.8SI161181.81063.8a Protocols include CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE* protocol refers to the modified CE protocol Table 4De novo assembled average contig numbers and N50 across protocols for Group A and CGroupProtocol aAverage number of contigsAverage N50 A CL155375.3891.3CE101836.5410.6TL129002.6821.3SI145,718686.4 C CE*136432.8842.6CP171464.0798.8SI161181.81063.8a Protocols include CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE* protocol refers to the modified CE protocol De novo assembled average contig numbers and N50 across protocols for Group A and C a Protocols include CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE* protocol refers to the modified CE protocol Table 5De novo assembled contigs identity composition across protocols for Group A and CProtocols aCLCETLSIAverageSt. deviationAverageSt. deviationAverageSt. deviationAverageSt. deviation Group A MVVirus10.6434.6404.2895.9724.4532.7125.9174.324Viroid0.0160.0090.0900.1440.0060.0030.0150.006Other b88.3344.71494.4166.23294.6722.73493.2524.500Unidentified1.0080.0751.2040.2900.8690.0620.8150.200MLVirus4.7542.1732.1212.3465.4233.6728.60410.462Viroid0.0030.0010.0030.0010.0030.0030.0040.003Other94.1032.07596.7572.73393.8103.73190.44210.657Unidentified1.1400.1171.1190.5060.7650.1480.9500.234 Protocols CE* CP SI AverageSt. deviationAverageSt. deviationAverageSt. deviation Group C MVVirus14.1542.60614.7603.72514.5255.630Viroid0.0360.0200.0370.0270.0140.007Other85.0772.54784.4873.73984.7485.557Unidentified0.7330.0710.7160.0080.7120.084a Protocols include CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE* protocol refers to the modified CE protocolb Refers to all other types of hits as well as host genome Table 5De novo assembled contigs identity composition across protocols for Group A and CProtocols aCLCETLSIAverageSt. deviationAverageSt. deviationAverageSt. deviationAverageSt. deviation Group A MVVirus10.6434.6404.2895.9724.4532.7125.9174.324Viroid0.0160.0090.0900.1440.0060.0030.0150.006Other b88.3344.71494.4166.23294.6722.73493.2524.500Unidentified1.0080.0751.2040.2900.8690.0620.8150.200MLVirus4.7542.1732.1212.3465.4233.6728.60410.462Viroid0.0030.0010.0030.0010.0030.0030.0040.003Other94.1032.07596.7572.73393.8103.73190.44210.657Unidentified1.1400.1171.1190.5060.7650.1480.9500.234 Protocols CE* CP SI AverageSt. deviationAverageSt. deviationAverageSt. deviation Group C MVVirus14.1542.60614.7603.72514.5255.630Viroid0.0360.0200.0370.0270.0140.007Other85.0772.54784.4873.73984.7485.557Unidentified0.7330.0710.7160.0080.7120.084a Protocols include CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE* protocol refers to the modified CE protocolb Refers to all other types of hits as well as host genome De novo assembled contigs identity composition across protocols for Group A and C a Protocols include CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE* protocol refers to the modified CE protocol b Refers to all other types of hits as well as host genome Reference gene mapping. All reads were mapped to twelve reference genes (Additional file 5) to assess the biological and technical variation associated with RNA extraction method in Group A and Group C. The TPM count of each gene was used to determine the reference gene expression profile associated with an extract (plant and extraction protocol combination) (Fig. 6). For both Madam Vinous and Mexican Lime samples in Group A, the CL, TL and SI extracts had similar profiles across biological replicates (Fig. 6a and c). The CE extracts, however, had considerable variation. Conversely, in Group C with the modified CE protocol extracts, the differences between expression profiles of samples were more measured, and this held true for the SI and CP extracts as well (Fig. 5e). In Group A and C, the gene with the lowest TPM count for all protocols was FBOX (Fig. 6a, c and e). The genes with the highest TPM count for CL and TL extracts in Group A and all extracts in Group C were TUB, EF-a1 and GAPC1 (Fig. 6a, c and e). For SI extracts in Group A, it was also TUB and EF-1a along with SAND. The CE extracts from Group A had more variation between Madam Vinous and Mexican Lime samples although TUB still had the highest TPM for both citrus types. Considering the PCA plots of genes, it was evident that some clustering did occur in Group A and C (Fig. 6b, d and f). While not as distinctive in Madam Vinous Group A samples, the CTAB buffered extraction method extracts tended to group together i.e. CL with CE in Group A (Fig. 6b and d) and CP with modified CE in Group C (Fig. 6f). The coverage of reference genes in Group A was quite similar across CL, TL and SI extracts with an average coverage of 95% (Fig. 7). CE extracts displayed varying coverage across different genes (33–98%). However, in Group C all tested protocols, modified CE, CP and SI, resulted in extracts with consistently high genome coverage with an average of 97.5% (Fig. 7). No statistically significant differences (p-value > 0.05) were observed in TPM count for any reference gene between the SI and TL extracts for both Madam Vinous and Mexican Lime samples in Group A and supported the clustering observed in the PCA plots (Fig. 6. b and d). This was also valid for the datasets from CE and CP protocols in Group C. The CE extracts resulted in significantly lower TPM counts (p-value < 0.05) for the PTB1 gene, compared to SI extracts for both citrus types including the data from Group C despite the modification made to the CE protocol of Group C samples. Furthermore, the only other gene-protocol comparison that had similar testing outcomes in Madam Vinous and Mexican Lime samples (Group A) included PTB1 when comparing CL with SI extracts, where SI samples had higher TPM counts, FBOX between CE and TL extracts, where TL samples had a higher portion of reads, and EF-1a also between CE and TL extracts, where TL samples had a higher TPM count (p-value > 0.05). When considering the overall results of Madam Vinous (Group A), Mexican Lime (Group A) and Group C only UBC21 had statistically similar TPM counts (p-value > 0.05) across all protocol comparisons (consistent results regardless of sample grouping). All other genes had at least one protocol comparison that showed significance in either citrus types or data grouping. Fig. 6Stacked column graphs and principal component analysis (PCA) plots of the transcripts per million (TPM) count of reference genes in a and b for Madam Vinous Group A (MV), in c and d for Group A Mexican Lime (ML) and in e and f for Group C samples. Results are grouped according to protocols as CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE protocol from Group C (e and f) refers to the modified CE protocol. The reference gene abbreviations are ACT2: Actin-2, TUB: Beta-Tubulin, UBC21: Ubiquitin-conjugating enzyme 21, Unknown: Unknown protein, FBOX: F-box family protein, DIM1: DIM1 homolog/YLS8, EF-1a: Elongation factor 1-alpha, SAND: SAND family protein, UBC9: Ubiquitin conjugating enzyme 9, UPL7: Ubiquitin-protein ligase 7, PTB1: Polypyrimidine tract-binding protein 1, GAPC1: Glyceraldehyde-3-phosphate dehydrogenase Stacked column graphs and principal component analysis (PCA) plots of the transcripts per million (TPM) count of reference genes in a and b for Madam Vinous Group A (MV), in c and d for Group A Mexican Lime (ML) and in e and f for Group C samples. Results are grouped according to protocols as CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE protocol from Group C (e and f) refers to the modified CE protocol. The reference gene abbreviations are ACT2: Actin-2, TUB: Beta-Tubulin, UBC21: Ubiquitin-conjugating enzyme 21, Unknown: Unknown protein, FBOX: F-box family protein, DIM1: DIM1 homolog/YLS8, EF-1a: Elongation factor 1-alpha, SAND: SAND family protein, UBC9: Ubiquitin conjugating enzyme 9, UPL7: Ubiquitin-protein ligase 7, PTB1: Polypyrimidine tract-binding protein 1, GAPC1: Glyceraldehyde-3-phosphate dehydrogenase Fig. 7Box and whisker plots of the sequence coverage of each reference gene across protocols in Group A and C. The boxplots represent the coverage distribution within a citrus type, Madam Vinous (MV) or Mexican Lime (ML) for Group A as well as Group C samples. Protocols include as CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE protocol from Group C refers to the modified CE protocol. The reference gene abbreviations are ACT2: Actin-2, TUB: Beta-Tubulin, UBC21: Ubiquitin-conjugating enzyme 21, Unknown: Unknown protein, FBOX: F-box family protein, DIM1: DIM1 homolog/YLS8, EF-1a: Elongation factor 1-alpha, SAND: SAND family protein, UBC9: Ubiquitin conjugating enzyme 9, UPL7: Ubiquitin-protein ligase 7, PTB1: Polypyrimidine tract-binding protein 1, GAPC1: Glyceraldehyde-3-phosphate dehydrogenase Box and whisker plots of the sequence coverage of each reference gene across protocols in Group A and C. The boxplots represent the coverage distribution within a citrus type, Madam Vinous (MV) or Mexican Lime (ML) for Group A as well as Group C samples. Protocols include as CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE protocol from Group C refers to the modified CE protocol. The reference gene abbreviations are ACT2: Actin-2, TUB: Beta-Tubulin, UBC21: Ubiquitin-conjugating enzyme 21, Unknown: Unknown protein, FBOX: F-box family protein, DIM1: DIM1 homolog/YLS8, EF-1a: Elongation factor 1-alpha, SAND: SAND family protein, UBC9: Ubiquitin conjugating enzyme 9, UPL7: Ubiquitin-protein ligase 7, PTB1: Polypyrimidine tract-binding protein 1, GAPC1: Glyceraldehyde-3-phosphate dehydrogenase CTV genotyping with read mapping CTV genotyping was performed on all samples using a strategy that employed stringent filtering of mapped reads to reference sequences representative of eleven CTV genotypes. Both TPM counts and genome coverage were used to assess the genotype content of each sample. The most abundant genotype in all samples was RB (Fig. 8) although T3 was also very dominant in four out of the five Mexican Lime samples (Fig. 8b). Variation was more evident between plants rather than between protocols which is indicative of biological variation as also indicated by RT-PCR genotyping results in Additional file 6. Overall, the CTV genotype profiles of plants were consistent across protocols. Only three genotypes; L1, M2 and T36; had an average coverage less than 50% for all samples (Fig. 9). RB, VT and T3 had more than 90% genome coverage across all extracts with RB and VT having the highest and most consistent coverage irrespective of protocol and sample (> 98%). All genotypes that were not present in samples as determined with RT-PCR (Additional file 6) had less than 90% coverage across all extracts, except T3. CL, TL, SI, CP and modified CE extracts from MV8, MV9, MV30 and ML27 had T3 coverage between 90 and 99% even though these plants had a negative result with RT-PCR (Fig. 9). Fig. 8Stacked column graphs of the transcripts per million (TPM) count of citrus tristeza virus (CTV) genotypes in a for Madam Vinous Group A (MV), b for Group A Mexican Lime (ML) and c for Group C samples. Samples are grouped according to extraction protocol. RNA extraction protocols are CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol) and SI (SDS/isopropanol) for Group A and modified CE, CP (CTAB/isopropanol) and SI for Group C Stacked column graphs of the transcripts per million (TPM) count of citrus tristeza virus (CTV) genotypes in a for Madam Vinous Group A (MV), b for Group A Mexican Lime (ML) and c for Group C samples. Samples are grouped according to extraction protocol. RNA extraction protocols are CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol) and SI (SDS/isopropanol) for Group A and modified CE, CP (CTAB/isopropanol) and SI for Group C Fig. 9Box and whisker plots of the genome coverage of each citrus tristeza virus (CTV) genotype across protocols in Group A and C. The boxplots represent the coverage distribution within a citrus type, Madam Vinous (MV) or Mexican Lime (ML) for Group A as well as Group C samples. RNA extraction protocols are CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol) and SI (SDS/isopropanol) for Group A and modified CE, CP (CTAB/isopropanol) and SI for Group C Box and whisker plots of the genome coverage of each citrus tristeza virus (CTV) genotype across protocols in Group A and C. The boxplots represent the coverage distribution within a citrus type, Madam Vinous (MV) or Mexican Lime (ML) for Group A as well as Group C samples. RNA extraction protocols are CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol) and SI (SDS/isopropanol) for Group A and modified CE, CP (CTAB/isopropanol) and SI for Group C Viruses and viroids read mapping To assess the pathogen profile and relative abundance associated with each extraction method, read mapping was performed using pathogen reference genomes corresponding to the full complement of the source plant: CTV, CiVA, CTLV, CDVd, CEVd, HSVd (Fig. 10). Based on the aforementioned genotyping results, eight CTV genotypes; RB, VT, T3, A18, HA16-5, S1, T30 and T68; were included in this mapping instance and the cumulative number of reads were used to assess CTV presence. In Madam Vinous and Mexican Lime samples of Group A, the CE extracts had the lowest CTV TPM count, although only significant for Madam Vinous (p < 0.05) (Fig. 10a). The other extracts; from CL, TL and SI protocols; were non-significantly different from one another. A similar result was obtained in Group C wherein the modified CE protocol, CP and SI extracts had indistinguishable differences. Mexican Lime samples from Group A had limited plants containing CTLV, CiVA and/or CEVd and the positive plants had relatively low concentrations which constrained assessment of these pathogens in this citrus type. Nonetheless, for CTLV in Madam Vinous Group A, the SI and TL extracts resulted in the highest TPM counts and was significantly more than the TPM counts obtained in the CE extracts (Fig. 10b). Both CiVA genomes were included in the read mapping and similar to CTLV, the SI and TL extracts of Madam Vinous Group A performed the best (highest TPM counts) for both genomes, but this was more pronounced for CiVA-RNA1 (Fig. 10c). For RNA2, the SI extracts had lower TPM counts compared to TL extracts (Fig. 10d). When considering CTLV and both CiVA genomes the extracts from Group C across all protocols had similar TPM counts (Fig. 10b and c). Across all data groupings, CE extracts had the highest TPM counts for all viroids (Fig. 10e-g) even though the CE protocol of Group C was modified. A likely consequence of this was that the performance of CE was inflated in Madam Vinous Group A and Mexican Lime samples compared to Group C. Although significant differences were mostly observed between CE and the protocol with the lowest TPM count in each case, it is important to note the magnitude of the differences between TPM counts as a limited sample number may have influenced statistical testing outcomes. Despite the modified CE protocol producing the highest TPM counts in Group C extracts, differences between protocols were more subtle compared to that observed in Group A. Genome coverage was high for pathogens in samples that tested positive in the initial screening (Fig. 11). Mexican Lime samples had an expanded coverage range due to the presence of samples that did not contain the full pathogen compliment, usually CiVA, CTLV and/or CEVd (according to screening), although sequencing results did show some indication of pathogen presence in some cases like low read counts and coverage. Most extracts in Madam Vinous Group A and Group C conferred genome coverage above 95% with some notable exceptions like CE extracts in Madam Vinous Group A that had lower coverage for CiVA-RNA1 and CTLV. The highest read depth for CTV was obtained with the CL and SI extracts (Fig. 12). SI extracts also had high read depth for the other viruses, CiVA and CTLV, along with TL and to a certain extent CL extracts. Irrespective of the modification made to the CE protocol, all CE extracts had high read depth for viroids closely followed by extracts from the other CTAB extraction buffer protocols CL and CP. The consensus sequences extracted from read mapping had no noticeable differences in sequence identity compared to the reference genome between protocols (Fig. 13). More than a 98% identity average was achieved in samples previously identified as positive for CEVd, HSVd, CiVA, CTLV, CTV RB and T3. Fig. 10Bar graph of the average transcripts per million (TPM) count associated with each protocol across samples in Group A-C in a for citrus tristeza virus (CTV), in b for citrus tatter leaf virus (CTLV), in c for citrus virus A-RNA1(CiVA-RNA1), in d for CiVA-RNA2, in e for citrus dwarfing viroid (CDVd), in f for citrus exocortis viroid (CEVd) and in g for hop stunt viroid (HSVd). Samples are grouped according to citrus type, Madam Vinous (MV) and Mexican Lime (ML), and experimental Group A or C. RNA extraction protocols are CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol) and SI (SDS/isopropanol) for Group A and modified CE, CP (CTAB/isopropanol) and SI for Group C. Note that CTV as indicated here is depicted as the cumulative value of all genotype TPM counts included in the read mapping analysis: A18, T30, T3, T68, HA16-5, RB, VT and S1. Letters a-c shown on top of the error bars indicate the statistical significance between protocols within a group. Similar letters are indicative of non-significant differences (p > 0.05) Bar graph of the average transcripts per million (TPM) count associated with each protocol across samples in Group A-C in a for citrus tristeza virus (CTV), in b for citrus tatter leaf virus (CTLV), in c for citrus virus A-RNA1(CiVA-RNA1), in d for CiVA-RNA2, in e for citrus dwarfing viroid (CDVd), in f for citrus exocortis viroid (CEVd) and in g for hop stunt viroid (HSVd). Samples are grouped according to citrus type, Madam Vinous (MV) and Mexican Lime (ML), and experimental Group A or C. RNA extraction protocols are CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol) and SI (SDS/isopropanol) for Group A and modified CE, CP (CTAB/isopropanol) and SI for Group C. Note that CTV as indicated here is depicted as the cumulative value of all genotype TPM counts included in the read mapping analysis: A18, T30, T3, T68, HA16-5, RB, VT and S1. Letters a-c shown on top of the error bars indicate the statistical significance between protocols within a group. Similar letters are indicative of non-significant differences (p > 0.05) Fig. 11Box and whisker plots of the genome coverage of pathogens across protocols in Group A and C. The boxplots represent the coverage distribution within a citrus type, Madam Vinous (MV) or Mexican Lime (ML) for Group A as well as Group C samples. Genome coverage results of extracts from positive plants (as determined by the initial screening) were used to construct these plots. Pathogens include citrus dwarfing viroid (CDVd), citrus exocortis viroid (CEVd), hop stunt viroid (HSVd), citrus virus A (CiVA) and citrus tatter leaf virus (CTLV). RNA extraction protocols are CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol) and SI (SDS/isopropanol) for Group A and modified CE, CP (CTAB/isopropanol) and SI for Group C. CTV genome coverage is not shown here Box and whisker plots of the genome coverage of pathogens across protocols in Group A and C. The boxplots represent the coverage distribution within a citrus type, Madam Vinous (MV) or Mexican Lime (ML) for Group A as well as Group C samples. Genome coverage results of extracts from positive plants (as determined by the initial screening) were used to construct these plots. Pathogens include citrus dwarfing viroid (CDVd), citrus exocortis viroid (CEVd), hop stunt viroid (HSVd), citrus virus A (CiVA) and citrus tatter leaf virus (CTLV). RNA extraction protocols are CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol) and SI (SDS/isopropanol) for Group A and modified CE, CP (CTAB/isopropanol) and SI for Group C. CTV genome coverage is not shown here Fig. 12Heatmap of the average read depth of pathogens across protocols in Group A and C. The heatmap is calculated for Group A: Mexican Lime samples, Group A: Madam Vinous samples and Group C separately. Orange indicates the minimum value, and dark green indicates the highest value. Pathogens include citrus dwarfing viroid (CDVd), citrus exocortis viroid (CEVd), hop stunt viroid (HSVd), citrus virus A (CiVA), citrus tatter leaf virus (CTLV) and citrus tristeza virus (CTV). RNA extraction protocols are CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol) and SI (SDS/isopropanol) for Group A and modified CE, CP (CTAB/isopropanol) and SI for Group C Heatmap of the average read depth of pathogens across protocols in Group A and C. The heatmap is calculated for Group A: Mexican Lime samples, Group A: Madam Vinous samples and Group C separately. Orange indicates the minimum value, and dark green indicates the highest value. Pathogens include citrus dwarfing viroid (CDVd), citrus exocortis viroid (CEVd), hop stunt viroid (HSVd), citrus virus A (CiVA), citrus tatter leaf virus (CTLV) and citrus tristeza virus (CTV). RNA extraction protocols are CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol) and SI (SDS/isopropanol) for Group A and modified CE, CP (CTAB/isopropanol) and SI for Group C Fig. 13Heatmap of the percentage sequence identity across protocols in Group A and C. The heatmap is calculated for the entire dataset ranging from 41.35–99.90%. Orange indicates the minimum value, and dark green indicates the highest value. Pathogens include citrus dwarfing viroid (CDVd), citrus exocortis viroid (CEVd), hop stunt viroid (HSVd), citrus virus A (CiVA), citrus tatter leaf virus (CTLV) and citrus tristeza virus (CTV). RNA extraction protocols are CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol) and SI (SDS/isopropanol) for Group A and modified CE, CP (CTAB/isopropanol) and SI for Group C Heatmap of the percentage sequence identity across protocols in Group A and C. The heatmap is calculated for the entire dataset ranging from 41.35–99.90%. Orange indicates the minimum value, and dark green indicates the highest value. Pathogens include citrus dwarfing viroid (CDVd), citrus exocortis viroid (CEVd), hop stunt viroid (HSVd), citrus virus A (CiVA), citrus tatter leaf virus (CTLV) and citrus tristeza virus (CTV). RNA extraction protocols are CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol) and SI (SDS/isopropanol) for Group A and modified CE, CP (CTAB/isopropanol) and SI for Group C Discussion This study evaluated different RNA extraction protocols in two citrus types using both RT-qPCR and HTS for the ability to detect citrus viruses and viroids. Madam Vinous and Mexican Lime plants were graft inoculated with a source plant positive for a range of pathogens and bark and leaf material was sampled from ten plants of each citrus type (Fig. 1). These samples were then subjected to four extraction protocols: CL, CE, TL and SI (Group A). RT-qPCR was then applied for CTV, CiVA, CDVd, CEVd, and HSVd quantitation in all extracts. These findings were then confirmed with five leaf samples (all extracts from each sample/plant) from each citrus type using Illumina HTS which enabled the evaluation of an additional pathogen, CTLV. The ten originally selected Madam Vinous plants’ leaves were resampled and subjected to CE and SI protocols in addition to the CP protocol (Group B) which was not included in the previous group. Pathogens were also quantitated in extracts using RT-qPCR. Five alternative Madam Vinous plants were then selected and subjected to the CE, SI and CP extraction protocols; however, the CE protocol was modified to not bias the extraction towards small RNAs (Group C). As with the previous groups, these extracts underwent RT-qPCR and all extracts from all samples were sent for HTS. Numerous nucleic acid extraction protocols and enrichment strategies have been developed to target specific pathogens based on defined characteristics; like genome type and size; and biological attributes; like replication intermediates and plant defence responses [1, 2, 16, 22, 31, 52]. While these protocols can significantly enhance pathogen detection of the defined group, it can often have a detrimental effect on broader pathogen detection [1]. Viroids in particular are elusive since the small size and unique secondary structures coupled with low concentration and uneven distribution [21] often associated with these pathogens, adversely affect many detection strategies. In the context of plant and crop protection strategies, an extraction protocol that facilitates broad range pathogen detection can be beneficial to increase efficiency of clean plant material generation and diagnostic testing. The question remains whether the possible adverse influence of an extraction method on one pathogen is outweighed by the benefits it provides to another. Some of the most common molecular assays like PCR and HTS are reliant on nucleic acid as input and optimisation of extraction protocols are essential to enhance the overall pathogen detection pipeline. The quality and purity of extracts can influence the overall performance of various detection assays and, therefore, contribute to the suitability of the protocol within a specific context. The purity of nucleic acids obtained with all protocols were sufficient except for the 260/230 ratio of CE (Table 1). However, when the modification was made to CE (Group C) there was considerable improvement observed. Typically, low 260/230 ratios are indicative of contaminants with absorbance at 230 nm [22], but since the original CE protocol had a severe deficit of rRNA and an abundance of small RNAs, it is possible that the increased rRNA obtained with the modification led to higher absorbance at 260 and RNA amount than what could be achieved with small RNAs alone. Additionally, a low amount of RNA (as achieved when only extracting small RNAs) might have low purity ratios because of the imbalance between amount and contaminants [18]. The modification to CE essentially minimised the exposure of nucleic acid to cold C:I, possibly improving the stability of RNA. Given the nature of the modification, it highlighted the variability of the protocol. RNA purity can influence pathogen detection and quantitation, although present findings indicated no observable debilitating effects on RT-qPCR efficiency for any extraction protocol. All the CE extracts were also flagged during HTS quality assessment for low RIN values and rRNA ratios (Table 2). The modification to the CE protocol improved both metrics. However, no considerable difference in HTS data generation was observed and data quality across protocols highlighted the robustness of Illumina sequencing. The amount of data, average quality scores and the data retention after quality trimming were similar for all extracts (Table 3). The CL protocol is the current HTS RNA extraction method used and is extensively validated [1, 2]. It produced impeccable sequencing data and although it was not necessarily significantly better for a specific aspect, it performed the best on every account of data quality. In terms of pathogen quantitation with RT-qPCR, there were clear distinctions in the ability of protocols to extract different pathogens. In Group A, the CL protocol outperformed all others regarding CTV extraction (Fig. 4a). The viroids, however, had higher copy numbers in CE samples which is in accordance with previous observations of ethanol precipitation and citrus viroids [2] (Fig. 4c-e). The CE extraction results indicated a severe bias towards small RNAs which, considering the small size of viroids [21], increased viroid RNA present in the extracts. Interestingly, CL also performed well for CDVd extraction despite previous studies indicating that LiCl precipitation decreases viroid presence [1, 2, 11]. CiVA copy numbers were higher for SI extracts which was also observed in Group C samples (Fig. 5b). Resampling of Madam Vinous plants (Group B) reflected initially obtained results regarding CE and SI protocols, although all pathogens’ copy numbers increased over time (seven-month sampling difference) as the pathogens continued to accumulate (Fig. 5). An increase in pathogen concentration over time (Group B) can have a positive effect by increasing the ease with which pathogens are detected, but earlier timepoints (Group A) can be more informative regarding the sensitivity of protocols and assays for early detection when concentrations are normally lower. The congruency in trends between Group A and Group B was, therefore, particularly encouraging that true variations due to protocol differences were being observed. The CP protocol performed better for virus detection than CE, but poorer compared to SI. When comparing SI extracts with the modified CE protocol and CP extracts (Group C), SI extracts yielded the highest CTV copy numbers similar to what was observed in Group A between CE and SI extracts (Fig. 5a). The same result was achieved with CiVA (Fig. 5b). Contrary to what was seen previously for viroids in Group A, the CE protocol did not perform the best for viroids in Group C due to the modification made (Fig. 5c-e). The modification was made to increase the abundance of larger RNA species in extracts to improve the performance of CE for virus detection, which it did achieve as the modified CE protocol performed similar to CP for all pathogens. In this instance the SI extracts had higher viroid RNA levels. The pathogen profiles as determined with both RT-qPCR and de novo assembly of HTS data had some contradictory results compared to initial screening of plants. This observation was mostly attributed to Mexican Lime samples regarding viroid amplification and assembly, but even some Madam Vinous samples had no assembled contigs corresponding to the HSVd reference sequence (Additional file 7). As previously mentioned, viroid detection is often impeded by the pathogen distribution and accumulation behaviour in plants [1, 21]. The assembled contigs’ N50 values were also considerably larger than average viroid size, although CE extracts in Group A were more similar to viroids (Table 4), which might be indicative of the decreased suitability and efficacy of viroid assembly. Only CL and SI extracts, from Madam Vinous Group A, and CP extracts from Group C had contig hits corresponding to the full pathogen compliment. The Mexican Lime dataset had at least one extract from each protocol that did not generate Blastn identifiable contigs corresponding to the profile indicated in the initial screening. Difficulties with viroid de novo assembly have been previously reported and consequently assembly is generally complemented by a read mapping approach [1]. Contig identity composition showed some variation across plants possibly alluding to the differential pathogen detection outcome observed with RT-PCR and Blastn. Overall, the average virus component across protocols was similar, except for CE extracts in Group A (Table 5). The viroid proportion, although relatively small, was the largest for CE extracts, even with the modified CE protocol. The modified CE protocol extracts had a larger viroid contig composition compared to SI extracts and a similar composition to CP extracts in Group C. Citrus reference genes were evaluated with a read mapping strategy to not only identify possible outlier samples, but also to assess which of the twelve were appropriate reference genes for cross citrus type and protocol comparison. The overarching trends observed in the TPM profiles were consistent across all but one protocol (Fig. 6). The CE protocol (as applicable to Group A) had considerable variation in gene TPM counts between plants and was also associated with lower and more varying genome coverage (Fig. 7). A slight grouping was observed in TPM counts for CTAB buffer protocols which indicated that for reference genes, differences were likely the cause of the buffer and associated protocol rather than precipitation method since ethanol precipitation yielded total nucleic acid with CTAB while LiCl merely had total RNA (Fig. 6b, d, f). Genes that had consistently high genome coverage across all protocols were ACT2, EF-1a and GAPC1, but the only gene that had no significant difference for any protocol comparison in all datasets (Madam Vinous Group A, Mexican Lime Group A and Group C) was UBC21. This highlighted the need to constantly validate reference genes within a specific context. Reference genes can serve as useful internal controls to normalise samples or to identify outliers. Outliers may indicate samples in which false negative detections could occur. The appropriate set of reference genes, however, may vary depending on the study parameters. All plants screened positive for CTV genotypes RB and VT which supported the consistently high genome coverage and TPM counts observed in sequencing data (Additional file 6 and Fig. 8). Genotypes L1, M2 and T68 were excluded in subsequent pathogen read mapping due to genome coverage below 50% which was indicative of non-target mapping [48]. Genotyping with HTS only contradicted T3 screening RT-PCR results but considering the sampling time difference between these two analyses (four months), it might be of biological consequence, or the PCR assays might have less sensitivity. Overall, the CTV genotype profiles of plants (Fig. 8) were consistent across protocols highlighting that different protocols do not necessarily favour certain genotypes but rather produce similar relative genotype proportions. The genotypes with the highest genome coverage also had the highest sequence identity compared to reference sequences (Fig. 13). CTV genotype composition appeared impervious to change in extraction protocol and the relative profiles were consistent in the same plant across extracts. As seen with RT-qPCR, there was some variation in pathogen abundance with HTS across protocol extracts. Notably, the CE protocol performed less ideally for virus detection but had the highest associated relative read counts (TPM) for all viroids compared to CL, SI and TL protocols which is in accordance with what was achieved with RT-qPCR (Fig. 10). SI and TL data had high TPM counts for CTLV and also CiVA, which in the case of CiVA were similar to pathogen quantitation results. Given the similar sizes of CiVA-RNA1 and CTLV [20, 53], results suggested that the SI and TL protocol were more inclined to extract RNA from this size range. Unlike RT-qPCR findings, the relative read counts for CTV were similar for CL, SI and TL. The SI, CP and the modified CE protocol, also resulted in similar TPM counts for nearly all pathogens. It is, however, important to note while not statistically significant, the modified CE protocol still had the highest viroid TPM counts which was in stark contrast to RT-qPCR where SI had considerably higher viroid copy numbers for Group C samples. No definitive differences were observed between protocols regarding genome coverage (Fig. 11) and read mapping consensus sequence identity (Fig. 13). The sequence identity for CDVd was consistently lower for all extracts compared to the other viroids and given the amount of ambiguous bases present in consensus sequences and the overall alignment of these sequences with the reference (data not shown) this might be a consequence of multiple viroid variants within certain plants. Depth of coverage appeared to be more linked to relative abundance as CE (including the modified CE) had higher depth for viroids but relative to other protocols, low depth for CTV (Fig. 12). A balance between practicality and detection capability needs to be struck for routine broad pathogen application. Of the protocols included in this study, SI and TL had relatively short experimental times while all CTAB protocols had longer waiting and centrifugation steps. Molecularly all protocols were able to repeatedly extract sufficient amounts of CTV RNA to enable detection and it is likely that most extraction protocols would be able to achieve this given the high levels of accumulation observed with CTV [1, 2]. The CE protocol showed some bias for small RNAs and consequently viroid RNA. The modification made to the CE protocol improved virus detection, and although a decreased amount of viroid RNA was supported by copy numbers and TPM counts being more comparable to other protocols, it did not debilitate the ability of the protocol to efficiently extract viroid RNA. Overall, the CE protocol was the only evaluated protocol that had an indication of detrimental effects on certain pathogens’ (CiVA and CTLV) detection. In the context of this study, it is important to prioritise reliability and consistency across all pathogens. Protocols like CL and SI performed well for overall pathogen detection without severe fluctuations between pathogens and pathogen bias like observed with CE. Ideally one protocol applicable to various pathogen detection pipelines would contribute greatly towards the streamlining and standardisation of routine diagnostics as well as “future proofing” for possible new technology developments. It is also well established that compounds or contaminants present in nucleic acid extracts can inhibit or compromise downstream analysis and the severity of this influence might differ between detection methods because some might be more robust than others [15, 22]. It is, therefore, imperative to assess the performance of the protocols across detection methodologies to determine suitability and characterise limitations. Conclusions This study showcased how the nucleic acids extracted can influence accurate pathogen detection across not only citrus types, but tissue types and detection methodologies. It investigated the influence of extraction protocols on multiple citrus viruses and viroids which enabled the evaluation of these pathogens simultaneously. In most cases, the expected pathogen profile could be confirmed with RT-qPCR and HTS irrespective of extraction protocol. De novo assembly of pathogens, especially viroids, were sometimes insufficient in plants where the concentration was low but coupled with read mapping the expected results were obtained. Results obtained with HTS were in accordance with RT-qPCR findings except for the performance of protocols in the additional comparison (Group C) for viroid quantitation, reiterating the importance of repeatability and validation across methodologies especially in the case of viroids where erratic detection behaviour is often observed. Overall, CL performed the best for CTV extraction while SI was the most sensitive for CiVA and CTLV. Viroid detection was more varied with both CL and CE yielding a high amount of CDVd RNA but for CEVd and HSVd the CE extracts had a higher abundance of these viroids. In the additional comparison of three protocols differences between protocols were milder but SI still had a better performance for virus extraction compared to CP and modified CE. While most protocols could accurately identify the target pathogens with RT-qPCR and HTS, the CE protocol was prone to severe fluctuations between pathogens exhibiting high viroid amounts and low CiVA and CTLV. CE extracts were also the least comparable to all other protocols’ extracts in terms of quality and the amount of pathogen copy numbers and TPM counts. The modification to the CE protocol did, however, enhance virus detection while maintaining adequate viroid quantities, although, this emphasised that even though the CE protocol can be effective, it is not a robust protocol. In the context of routine broad range pathogen detection capabilities, the possible advantageous extraction of one pathogen does not necessarily outweigh the detrimental influence it may have on others. Consistency and performance must be carefully evaluated and regardless of what protocol is used, the proper characterisation of the associated limitations is essential. Being aware of the possible restrictions of any protocol can greatly enhance the performance thereof because compensations can be made to mitigate the extent of these influences. Present findings indicate that the CL and SI protocols are the best overall extraction protocols for broad citrus virus and viroid detection across both RT-qPCR and HTS. The CL protocol showed the most consistent high-quality performance while the SI protocol offered a practical, cost-effective, and efficient alternative. The CE protocol, especially unmodified, is not suitable for routine broad-range diagnostics due to severe bias toward viroids and less consistent virus detection, and while the TL and CP protocols were acceptable, they do not offer any major advantage over the CL or SI protocols. These results highlight the need for constant re-evaluation, improvement and optimisation to ensure that the most appropriate methods are employed for routine implementation in certification and disease management programs. Constant vigilance of these programs is essential to ensure the prosperity of not only the local but global citrus industry. Supplementary Information Supplementary Material 1: RT-PCR assay specifications for screening of citrus tristeza virus, citrus virus A, citrus dwarfing viroid, citrus exocortis viroidand Supplementary Material 1: RT-PCR assay specifications for screening of citrus tristeza virus, citrus virus A, citrus dwarfing viroid, citrus exocortis viroidand Supplementary Material 2: RT-PCR assay specifications for citrus tristeza virus (CTV) genotyping. Supplementary Material 2: RT-PCR assay specifications for citrus tristeza virus (CTV) genotyping. Supplementary Material 3: gBlock Gene Fragments sequences used for absolute quantitation. Pathogens for which gBlocks were designed include citrus tristeza virus (CTV), citrus virus A (CiVA), citrus dwarfing viroid (CDVd), citrus exocortis viroid (CEVd) and (HSVd) Supplementary Material 3: gBlock Gene Fragments sequences used for absolute quantitation. Pathogens for which gBlocks were designed include citrus tristeza virus (CTV), citrus virus A (CiVA), citrus dwarfing viroid (CDVd), citrus exocortis viroid (CEVd) and (HSVd) Supplementary Material 4: Specific primers and cycling conditions used in the standard curve qPCR assays and RT-qPCR assays of RNA extracts. Pathogens indicated include citrus dwarfing viroid (CDVd), citrus exocortis viroid (CEVd), hop stunt viroid (HSVd), citrus tristeza virus (CTV), citrus tatter leaf virus (CTLV) and citrus virus A (CiVA). CRI – Citrus Research International Supplementary Material 4: Specific primers and cycling conditions used in the standard curve qPCR assays and RT-qPCR assays of RNA extracts. Pathogens indicated include citrus dwarfing viroid (CDVd), citrus exocortis viroid (CEVd), hop stunt viroid (HSVd), citrus tristeza virus (CTV), citrus tatter leaf virus (CTLV) and citrus virus A (CiVA). CRI – Citrus Research International Supplementary Material 5: Reference sequences utilised in read mapping analyses. Supplementary Material 5: Reference sequences utilised in read mapping analyses. Supplementary Material 6: Pathogen screening and citrus tristeza virus (CTV) genotyping results of selected plants. Pathogens indicated include citrus dwarfing viroid (CDVd), citrus exocortis viroid (CEVd), hop stunt viroid (HSVd), citrus tristeza virus (CTV), citrus tatter leaf virus (CTLV) and citrus virus A (CiVA). Citrus types include Madam Vinous (MV) and Mexican Lime (ML). Supplementary Material 6: Pathogen screening and citrus tristeza virus (CTV) genotyping results of selected plants. Pathogens indicated include citrus dwarfing viroid (CDVd), citrus exocortis viroid (CEVd), hop stunt viroid (HSVd), citrus tristeza virus (CTV), citrus tatter leaf virus (CTLV) and citrus virus A (CiVA). Citrus types include Madam Vinous (MV) and Mexican Lime (ML). Supplementary Material 7: Pathogen detection results using de novo assembled contigs and Blastn. Results are grouped according to protocols as CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE* protocol refers to the modified CE protocol. Pathogens indicated include citrus dwarfing viroid (CDVd), citrus exocortis viroid (CEVd), hop stunt viroid (HSVd), citrus tristeza virus (CTV), citrus tatter leaf virus (CTLV) and citrus virus A (CiVA). Citrus types include Madam Vinous (MV) and Mexican Lime (ML). Supplementary Material 7: Pathogen detection results using de novo assembled contigs and Blastn. Results are grouped according to protocols as CL (CTAB/LiCl), CE (CTAB/ethanol), TL (acid phenol buffer/LiCl and isopropanol), CP (CTAB/isopropanol) and SI (SDS/isopropanol). The CE* protocol refers to the modified CE protocol. Pathogens indicated include citrus dwarfing viroid (CDVd), citrus exocortis viroid (CEVd), hop stunt viroid (HSVd), citrus tristeza virus (CTV), citrus tatter leaf virus (CTLV) and citrus virus A (CiVA). Citrus types include Madam Vinous (MV) and Mexican Lime (ML).