Artificial Intelligence-Assisted Loop Mediated Isothermal Amplification (ai-LAMP) for Rapid and Reliable Detection of SARS-CoV-2

Until vaccines and effective therapeutics become available, the practical solution 27 to transit safely out of the current coronavirus disease 19 (CoVID-19) lockdown may include 28 the implementation of an effective testing, tracing and tracking system. However, this 29 requires a reliable and clinically validated diagnostic platform for the sensitive and specific 30 identification of SARS-CoV-2. Here, we report on the development of a de novo, high-31 resolution and comparative genomics guided reverse-transcribed loop-mediated 32 isothermal amplification (LAMP) assay. To further enhance the assay performance and to 33 remove any subjectivity associated with operator interpretation of results, we engineered a 34 novel hand-held smart diagnostic device. The robust diagnostic device was further 35 furnished with automated image acquisition and processing algorithms, and the collated 36 data was processed through artificial intelligence (AI) pipelines to further reduce the assay 37 run time and the subjectivity of the colorimetric LAMP detection. This advanced AI 38 algorithm-implemented LAMP (ai-LAMP) assay, targeting the RNA-dependent RNA 39 polymerase gene, showed high analytical sensitivity and specificity for SARS-CoV-2. A 40 total of ~200 coronavirus disease (CoVID-19)-suspected NHS patient samples were tested 41 using the platform and it was shown to be reliable, highly specific and significantly more 42 sensitive than the current gold standard qRT-PCR. Therefore, system could provide an 43 efficient and cost-effective platform to detect SARS-CoV-2 in resource-limited laboratories. 44

SARS-CoV-2, diagnosis, LAMP, point of care, artificial intelligence 45 46 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 15, 2020. ; https://doi.org/10.1101/2020.07.08.20148999doi: medRxiv preprint NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice. 1. Introduction 47 A cluster of new pneumonia cases was reported to the World Health Organization 48 (WHO) in late 2019 from Wuhan, Hubei Province of China. The causative agent was named 49 as severe acute respiratory syndrome coronavirus 2 (SARS -CoV-2) and led to a global 50 pandemic [1–3]. While the major impact of SARS -CoV-2 was attributed to frail and elderly 51 people with co -morbidities, coronavirus disease 2019 (CoVID-19) was mainly spread by 52 asymptomatic or mildly symptomatic patients [2]. Due to their high mutation rates and 53 recombination events, coronaviruses can infect a range of animal species including humans, 54 avian, rodents, carnivores, chiropters and oth er mammals [4]. Prior to the emergence of 55 SARS-CoV-2, a total of six different coronaviruses had been reported to infect humans, 56 including HCoV-229E, HCoV-OC43, HCoV-NL63, HCoV-HKU1, MERS and SARS -CoV-1 57 (also known as classical SARS). The SARS-CoV-2 belongs to the β-coronavirus of the group 58 2B within the family of Coronaviridae [3]. 59 The SARS -CoV-2 shares a high level of genetic similarity (up to 96%) with 60 coronaviruses originating from bats [3]. The genome of β -coronavirus encodes for the 61 replicase complex (ORF1ab), spike (S), envelope (E), membrane (M) and nucleoprotein (N) 62 genes in addition to the several non-structural and accessory proteins in the order from 5’ -63 untranslated to 3’-untranslated regions [3]. Owing to the nature of viral genetics, the N gene 64 is the most transcribed and highly conserved gene within the Coronaviridae family and has 65 been a major target for both antigen and antibodies diagnostics. Across the genome, the 66 RNA-dependent RNA polymerase (RdRP), encoded by the ORF1b gene segment, presents 67 a high level of intra -group conservation and therefore is an ideal target for a diagnostic 68 application [5, 6]. 69 As evident by previous coronaviruses associated pandemics and other viral diseases, 70 a highly specific, sensitive and easily deployable diagnostic is critical for the identification, 71 tracing, rationalizing control measures, documentation of symptomatic and asymptomatic 72 carriers [7-12]. Additionally, due to the unavailability of the regis tered vaccines or effective 73 therapeutics, rapid and reliable diagnostics are of paramount importance to curtail SARS -74 CoV-2 infection. Because of shortcomings associated with the virus isolation (time 75 consuming and required containment) and cross -reactivities of antigen and antibodies 76 assay, several real-time reverse transcription-polymerase chain reactions (qRT -PCR) and 77 reverse-transcription loop mediated isothermal amplification (RT-LAMP) assays have been 78 developed, validated and commercialized as useful laboratory diagnostics for the detection 79 of SARS-CoV-2 [13]. However, the majority of these assays are time-consuming and require 80 laboratory-intense instrumentation. Furthermore, they are unable to meet the current 81 unprecedented rapid growth and demand for testing a large proportion of the population, 82 identification of asymptomatic carriers and contact tracing. 83 Though qRT-PCR remains the gold standard for the diagnosis of SARS -CoV-2, RT-84 LAMP assays have been demonstrated to produce diagnostic results with in creased 85 sensitivity and specificity [14]. Furthermore, its ability to tolerate PCR inhibitors eliminates 86 the need for laborious RNA extraction and purification methodologies [15, 1 6]. Several 87 platforms capable of performing LAMP assays in the field have previously been documented 88 [17]. However, most platforms have employed fluorescence detection with integrated optical 89 units or a smart phone dock to achieve detection [18, 19]. Similarly, for colorimetric LAMP 90 assays, smart phone cameras or user interpretati on of the colour changes were used to 91 achieve detection [20, 21]. The fully integrated real -time fluorescence-based platforms are 92 expensive, and the smartphone-based platforms are only designed for specific smartphone 93 models. Therefore, to fulfil the need for a standalone colorimetric isothermal nucleic acid 94 amplification platform [22], we have developed an ultra-low-cost molecular diagnostic device 95 with an integrated single -board computer, imaging camera, artificial intelligence -based 96 image processing algorithm and mobile app. 97 In this study, we developed a high -resolution comparative genomics analysis -guided 98 novel RT-LAMP assay for the specific and sensitive detection of SARS-CoV-2 in comparison 99 to WHO recommended qRT -PCR assays. In order to provide a s imple “sample-to-answer 100 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 15, 2020. ; https://doi.org/10.1101/2020.07.08.20148999doi: medRxiv preprint workflow”, an ultra -low-cost and user -friendly diagnostic platform was engineered and 101 further enhanced with a module for automated image acquisition and processing. Artificial 102 intelligence-guided assessment of the LAMP assay provide d faster detection of colour 103 changes in the LAMP reaction, further enhancing the assay performance and thus reducing 104 the potential for human error in results interpretation. Finally, the assay was validated on 105 RNA extracted from clinical samples from SARS-CoV-2 suspected patients to demonstrate 106 the real-life applicability. 107 2. Experimental Section 108 2.1. Ethics statement 109 This study was conducted in accordance with the University Human ethics guidelines 110 and received a favourable review from the Faculty of Health and Medicine Research Ethics 111 Committee (FHMREC) of Lancaster University - reference number FHMREC19112. The 112 study was exempt from requiring specific patient consent, as it only involved the use of 113 extracted RNA and existing collections of data or records that contained non -identifiable 114 data about human patients. 115 116 2.2. Cells and viruses 117 Vero cells and MDCK cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) 118 (Gibco, Carlsbad, CA) supplemented with 10% inactivated foetal bovine serum (FBS) 119 (Gibco), 2 mM l-glutamine (Gibco) and 100U/mL penicillin/streptomycin (Gibco) at 37°C in 120 5% CO 2. Influenza A virus (A/chicken/Pakistan/UDL-01/2008(H9N2), Newcastle disease 121 virus strain LaSota and Infectious bronchitis virus strain H120, Vesicular stomatitis virus 122 (VSV) and Sendai virus (SeV) were propagated and used to determine the specificity of the 123 LAMP. All viruses except influenza were titrated on Vero and MDCK cells, respectively by 124 the standard plaque assay. 125 126 2.3. In silico nucleotide sequence comparisons and primer design 127 To design specific LAMP primer sets for the detection of SARS -CoV-2, all available 128 complete genome sequences were downloaded from GISAID Initiative 129 (https://www.gisaid.org/), aligned and the conserved part was selected and used as the 130 template of LAMP primer design. To find out an efficient primer set, three sets of specific 131 LAMP primers were hand-picked and validated using PrimerExplorer V5 software 132 (http://primerexplorer.jp/elamp4.0.0/index.html). Primers were validated using BLAST 133 software (http://www.ncbi.nlm.gov/BLAST) to ensure their specificity. 134 135 2.4. Cloning and in vitro transcription of RdRP target gene 136 The coding sequence of SARS -CoV-2 RdRp gene was chemically synthesized and 137 cloned into pVAX1 plasmid (Invitrogen, Carlsbad, USA) between KpnI and NotI restriction 138 sites. The plasmid was propagated in DH5α cells and purified using MiniPrep Qiagen Kits. 139 The linearized plasmid with pVAX1-RdRP was used for in vitro transcription using T7 140 RiboMAX™ Express Large-Scale RNA Production System (Promega, USA). The copy 141 number of in vitro transcribed RNA was calculated from RNA concentration measured with 142 NanoDrop™ 2000c Spectrophotometers (Thermo, USA) in triplicate. RNA products were 143 then purified using the RNeasy MinElute Cleanup Kit (Qiagen, Valencia, CA, USA). A 144 standard curve was generated using dilutions of the standard in vitro transcribed RNAs using 145 SuperScript III Platinum One -Step qRT -PCR Kit as per the manufacturer’s protocol 146 (Invitrogen, Carlsbad, USA) using CFX384 Touch Real-Time PCR Detection System is 147 (Applied Biosystems, USA). 148 149 2.5. Clinical sample processing and spiking with miR-cel-miR-39-3p RNA 150 A total of 199 nasopharyngeal swabs were individually collected from CoVID-19 151 suspected patients, through routine NHS collection procedure for COVID-19 screening. 152 These samples were stored and transported in the virus transport media (VTM) to the NHS 153 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 15, 2020. ; https://doi.org/10.1101/2020.07.08.20148999doi: medRxiv preprint diagnostic laboratory at Lancaster University, UK. All samples were individually spiked with 154 50 pmol/L of synthesized Caenorhabditis elegans miR-cel-miR-39-3p (Applied 155 Biosystems/Thermo-Fisher Scientific, UK). The miR -cel-miR-39-3p RNA lacked any 156 sequence homology to human or viral gene and thus present an ideal RNA extraction 157 control. Total RNA including miRNAs was extracted using 140 μL of the spiked-VTM by the 158 commercial QIAampViral RNA Mini kit (Qiagen, Valencia, California). The miR-cel-miR-39-159 3p RNA was used to serve as an internal control to monitor extraction efficiency and used 160 for data normalization. The final RNA yield and purity were determined by the A260/A280 161 ratio measured by a NanoDrop ND-1000 spectrophotometer (NanoDrop 162 Technologies/Thermo-Fisher Scientific, UK) with a ratio of 1.80 to 2.00 indicative of good 163 RNA purity. The isolated RNA was stored at −80 °C for further use. 164 165 2.6. Real-time fluorescent-based quantitative PCR 166 Suspected SARS-CoV-2 clinical samples were tested for positivity by qRT-PCR. Briefly, 167 RNA was extracted from Viral Transport Media using the QIAamp Viral RNA Mini kit (Qiagen, 168 Valencia, California) following the manufacturer instructions. The qRT-PCR was conducted 169 using the SuperScript III Platinum One-Step qRT-PCR Kit as per the manufacturer’s protocol 170 (Invitrogen Carlsbad, USA) in the CFX384 Touch Real-Time PCR Detection System (Biorad, 171 USA), according to the cycling protocol. The reaction was performed using the s pecific 172 primer set RdRpF; RdRpR and FAM-labelled probe or NP-F; NP-R and ROX labelled probes 173 designed to detect SARS -CoV2. The 25-µl PCR reaction consists of 12.5 µl 2X Reaction 174 Mix, 0.2 µM of each primer, and 0.1 µM probe, 0.5 µl of SuperScript® III RT/Platinum® Taq 175 Mix, 5 µl of RNA sample and nuclear free water. The cycling program was performed in the 176 CFX384 Touch Real-Time PCR Detection System is (Applied Biosystems, USA), according 177 to the cycling protocol. The amount of viral RNA in each sample was estimated by comparing 178 the cycle threshold values (Ct) to the standard curve made by serial 10-fold serial dilutions 179 of previously titrated in vitro transcribed RdRP gene. 180 181 2.7. ai-LAMP assay performance 182 All experiments for LAMP were run in triplicate. The LAMP reactions were performed 183 using WarmStartTM Colorimetric LAMP 2X Master Mix (New England Biolabs). A 10X primer 184 mix (FIP, 16 µM; BIP, 16 µM; F3, 2 µM; B3, 2 µM; LF, 4 µM; LB, 4 µM) was prepared. A 25 185 µl reaction mixture (12.5 µl 2X MasterMix; 2.5 µl 10X primer mix; 2.5 µl RNA and 7.5 µl 186 DNase & RNase-free molecular grade water) was mixed homogeneously and centrifuged. 187 The LAMP assays were performed in a thermocycler (MJResearch) at 65°C for 30 min or in 188 the engineered device (Figure 4A). Colour change was observed directly by the naked eye 189 or through AI image processing, and agarose gel electrophoresis was performed to confirm 190 the results. The completion of amplification was indicated by the colour in the tube, wherein 191 yellow was considered positive and pink was regarded as negative. All amplicons were 192 confirmed by 2% agarose gel electrophoresis. 193 194 2.8. Artificial intelligence based test-tube colour detection 195 A loop-mediated isothermal amplification (LAMP) assay based COVID-19 test device 196 was developed to capture the COVID-19 test results in 30 minutes, based on colour 197 changes. Artificial intelligence (AI) based colour detection was proposed to identify colour 198 changes considering different lighting issues and to reduce the test running time less than 199 30 minutes. Images were acquired from the COVID-19 test kit which carried 8 test-tubes 200 including NTC (negative test control) and PTC (positive test control) for every 20 seconds 201 during the test operation. Each image was cropped into separate tubes using template 202 matching approach and labelled manually based on their colour. 203 204 2.9. Analytical specificity and analytical sensitivity of the LAMP assay 205 The designed RdRp primer sets for LAMP to detect SARS-CoV-2 were validated for 206 their specificity by testing the cross-reactivity with other viruses, including influenza A virus, 207 Vesicular stomatitis virus (VSV), Sendai virus (SeD), infectious bronchitis virus (IBV) and 208 Newcastle disease virus (NDV). Likewise, the developed LAMP assay was evaluated to test 209 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 15, 2020. ; https://doi.org/10.1101/2020.07.08.20148999doi: medRxiv preprint the primers set sensitivity in a serially diluted standard RNA template prepared by tenfold 210 serial dilutions. The amplification patterns were observed for each dilution to determine the 211 lowest amount of absolute RNA template required for detectable amplification. The degree 212 of colour intensity of the amplified product as well as the observed electrophoretic pattern 213 during gel electrophoresis was used for the analysis of LAMP amplification. 214 215 2.10. Quantitative real time PCR for miR-cel-miR-39-3p RNA 216 In order to determine the RNA extraction efficiency, the extracted RNA was reverse 217 transcribed using a commercially available kit (Applied Biosystems/Thermo -Fisher 218 Scientific, UK) using miR-specific stem-loop primers as per manufacturer instructions. A total 219 of 5 μL of the sample was added to a 96 -well plate together with 10 μL reaction mixture 220 (MasterMixTM) containing along with MultiscribeTM reverse transcriptase (50 U/μL), and 221 0.19 μL RNAase inhibitor (20 U/μL). The RT reaction was performed at 16 °C for 30 min, 222 followed by 42 °C for 30 min, and 85 °C for 5 min and was finally kept at 4 °C. A NTC was 223 considered in every individually run reaction to identify any unspecific amplification. The RT 224 products were quantified immediately by qPCR using TaqManTM Mi croRNA assays 225 (Applied Biosystems/Thermo-Fisher Scientific, UK) in a 96 well plate using the 7900HT Fast 226 Real-Time PCR System (Applied Biosystems, UK) as we described before [23]. The 227 quantification cycle (Cq) was determined with instrument default threshold settings (10 SDs 228 above the mean fluorescence of the baseline cycle). 229 230 2.11. Statistical analysis 231 A total of 200 sample size was calculated to assess the performance of the LAMP assay. 232 GraphPad Prism Software version 6.01 for Mac (GraphPad Software, La Jolla, California, 233 USA) was used for graphs generation. The LAMP detection sensitivity and specificity were 234 calculated using the chi -squared test. TPR (true positive rate), TNR (true negative rate), 235 FPR (false positive rate), FNR (false negative rate) were calculated according to the 236 following equations: TPR= TP/(TP+FN). TNR=TN/(FP+TN). FNR=FN/(TP+FN). 237 FPR=FP/(FP+TN). TP: total number of true positives. TN: total number of true negatives. 238 TN: total number of true negatives. FN: total number of false negatives. 239 3. Results 240 3.1. High resolution conversation analysis of SARS-CoV-2 to guide oligo design 241 It is imperative to critically assess the evolving nature of viruses in identifying conserved 242 gene signatures and guiding the selection of the most appropriate primers. In order to 243 identify important genomic loci, we downloaded and aligned all the available full -length 244 genomes with high coverage sequences ( n=22858) of SARS-CoV-2 by Multiple Alignment 245 using Fast Fourier Transform (MAFFT) [24]. We then compiled in house R-code (available 246 on request) to determine the single nucleotide-based genetic conservation across the length 247 of ~30kb genome. The analysis of the aligned dataset of all genomes in the RStudio 248 generated a total of 18GB high-resolution nucleotide-by-nucleotide score from 0.0 to 1.0 (1.0 249 being the highly conserved and 0.0 being the highly divergent). Plotting the assessed genetic 250 divergence, at a cut-off point of 90% similarity along with the genome of the SARS -CoV-2, 251 identified sharp diverg ence at multiple loci (Figure 1A). However, most of the genomes 252 maintained high conservation. The divergence at the 5' and 3' ends was primarily due to 253 length heterogeneity, which may be partly as a result of sequencing artifact or potentially 254 coronaviruses ragged termini (Figure 1B). Owing to high divergence, a stretch of sequence 255 (~400 nucleotides, numbering corresponds to the complete genome of strain SARS -CoV-256 2/human/USA/VA-DCLS-0285/2020 strain, GenBank Accession Number: MT558705.1) 257 spanning the start of the ORF1b, which encode for viral RNA -dependent RNA polymerase 258 (RdRP), was targeted to design oligos for the LAMP assay. Additionally, this specific target 259 genomic locus was adjacent to oligos recommended by the World Health Organization 260 (WHO) and Public Health England (PHE) for real-time RT-PCR-based routine identification 261 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 15, 2020. ; https://doi.org/10.1101/2020.07.08.20148999doi: medRxiv preprint of CoVID-19 patients, further allowing direct and comparable evaluation of real -time RT-262 PCR with de novo developed LAMP assay (Figure 1C). 263 The conserved region of the RdRP gene with the lowest mutation frequencies was used 264 as a template to manually design three sets of basic LAMP primers and selected with 265 PrimerExplorer V5 for appropriate primer lengths, loop selection and melting temperature 266 optimization (Figure 1C). In order to precl ude the non -specific amplification of common 267 coronaviruses, efforts were made to design primers in the regions where there is a high level 268 of divergence among more than 3 of the 6 total primers in a specific set. Amongst the most 269 suitable targets, the prim ers with high scores were aligned with MERS -CoV, hCoV-229E, 270 hCoV-OC43, hCoV -NL63, hCoV -HKU1 and SARS -CoV-1 (Figure 1D). These selected 271 primers were used for further validation and screening. 272 273 Figure 1. In silico analysis of SARS-CoV-2 and primer design. (A) Genome organization 274 of SARS -CoV-2. Scale represents an approximate position of the genome whereas 275 ORF1a and b are expanded to show internal gene organization. ( B) Level of gene 276 identity across the genome of the SARS -CoV-2. Identity less than 90% is not shown. 277 (C) Primer location in the RdRP gene of SARS -COV-2 is shown. Red coloured 278 sequences represent LAMP primers whereas blue coloured sequences are primers and 279 probes used in the qRT -PCR. (D) Comparative sequence identity using the primers 280 against different human coronaviruses compared to the reference SARS -CoV-2 281 sequence; dots represent identical nucleotides. 282 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 15, 2020. ; https://doi.org/10.1101/2020.07.08.20148999doi: medRxiv preprint 3.2. Determination of the limit of detection of the LAMP assay using biochemically 283 synthesised RNA 284 In order to assess the robustness of the primers, we used a fully identical in vitro 285 transcribed target RNA unanimously spanning the length of the RdRP -gene based LAMP 286 and qRT -PCR target regions. The pre -determined copy nu mbers of the biochemically 287 synthesised RNA were 10-fold serially diluted from 107 copies to 0 copies of the target gene 288 per reaction. To determine the analytical sensitivity of the assays, we first evaluated their 289 limits of detection (LoD) for both qRT -PCR and LAMP assays. The LoD of the qRT -PCR 290 was 10 copies as evident from the relati ve fluorescence units (Figure 2A) and 291 electrophoreses of the amplified products (Figure 2B). The standard curve generated by the 292 RdRP-based qRT-PCR was linear and generated a coefficient of correlation (R 2) = 0.9481 293 and a slope of -2.6509 (Figure 2C). Melting curve analysis revealed the specificity of primers 294 for the target gene sequence, as all the amplified products showed a uniform melting 295 temperature (Tm) of ~75.10°C and specific amplification patterns (Figure 2B and data not 296 shown). Compared to the qRT-PCR assay, the LoD for the LAMP which targeted the same 297 RdRP gene was 1 log unit higher (102 copies/reaction) (Figure 2D, upper panel) as assessed 298 by visual observation of the LAMP reaction, where positive reactions turned yellow and 299 negative reactions remained pink when observed by the naked eye. To further confirm the 300 specific amplification of the target region, the gradient LAMP products were visualized by 301 DNA staining and gel electrophoresis for the amplified product, further confirming the 302 detection limit of LAMP (Figure 2D, lower panel). 303 304 Figure 2: Sensitivities of the LAMP assay. ( A) Seven different dilutions of in vitro 305 transcribed RNA were run for quantitative measurement using qRT -PCR. Relative 306 fluorescence units show a gradient decrease in signals. ( B) The corresponding PCR 307 products on the electrophoresis gel (C) The qRT-PCR standard curve based on the Ct 308 value and dilution factor. (D) The serially diluted synthetic RNAs were run in the LAMP 309 assay and colour change represents positive (yellow) or negative (pink). The lower 310 panel shows the LAMP gradient products. 311 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 15, 2020. ; https://doi.org/10.1101/2020.07.08.20148999doi: medRxiv preprint 3.3. Cross-reactivity of the novel LAMP assay when tested against other respiratory and 312 medically important viruses 313 The SARS-CoV-2 embraces genetic and phenotypic features of several common cold 314 coronaviruses and other viruses of the respiratory tract. Owing to high genetic similarity (up 315 to 96% at nucleotide levels) and common respiratory specimen for clinical identification of 316 CoVID-19 patients, we aimed to investigate any non -specific amplification in the LAMP 317 assay. In order to demonstrate the specificity of the LAMP assay, we used pathogens 318 belonging to 5 families of the most important medical and respiratory viruses. As shown in 319 the Figure 3A, the qRT -PCR specifically detected only the SARS -CoV-2 and this was 320 confirmed by Gel -red staining of amplified products (Figure 3B). Consistently, no cross -321 reactivities were observed with the LAMP in both colorimetric detection (Figure 3C, upper 322 panel) or electrophoreses (Figure 3C, lower panel). Collectively, a highly specif ic detection 323 of SARS-CoV-2 was observed for primers set using either of the assays. 324 325 326 Figure 3: Specificity of the LAMP assay. (A) RNA extracted from different medically 327 or respiratory important viruses as well as two dilutions of synthetic RNA were run 328 for qPCR. ( B) Corresponding PCR products were run on gel to demonstrate 329 specificity. (C) Similar to qRT-PCR, extracted RNA were run in the LAMP assays. 330 The top panel indicates the colorimetric detection of LAMP positive/negative 331 reactions and the lower panel show the electrophoresis of the corresponding LAMP 332 products. 333 3.4. Temporal investigations of the LAMP assay and its impact on the limit of detection 334 One of the major advantages of LAMP is its robustness. In order to determine the o ptimal 335 time required for sufficient amplification of targeted genes, in vitro transcribed RNA was used 336 as a template in 30 minutes assays and assessed every 5 minutes post-start of the reaction. 337 The change in colour was monitored visually by the naked eye. As shown in Table 1, under 338 equivalent conditions similar results were obtained from between 20 -30 minutes of 339 amplification. Therefore, to improve sensitivity and avoid missing any weak positives, 30 min 340 was selected as the optimal visual interpretation time for the results. 341 While the change in colour, reflective of a positive reaction, could be detected as early 342 as 20 minutes post-start of the reaction at lower copy number, subjective variabilities may 343

in erroneous interpretation, especially in col orimetric based diagnostic assays. To 344 propose an automated imaging, processing and interpretation of the LAMP based results, 345 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 15, 2020. ; https://doi.org/10.1101/2020.07.08.20148999doi: medRxiv preprint we developed a user -friendly device and furnished it with an artificial intelligence based 346 automatic interpretation algorithm. 347 348 349 Table 1. Reaction times to visually detect LAMP positivity 350 Time in minutes (after start of the LAMP reaction) In vitro transcribed RNA dilution (copies/reaction) 107 106 105 104 103 102 10 0 NTC 05 - - - - - - - - - 10 - - - - - - - - - 15 + + + + - - - - - 20 + + + + + + - - - 25 + + + + + + - - - 30 + + + + + + - - - 351 3.5. Manufacture of an isothermal nucleic acid amplification device with colorimetric 352 detection features 353 A device (Figure 4A) was built with many off-the-shelf electronic components and 354 custom flexible resistive heating elements (5W, NEL, UK), and specially designed aluminium 355 heating blocks. Raspberry Pi (RPi) was used to control the device. The one wire interface of 356 the RPi was used to connect ten digital temperature sensors (DS18B20, Maxim Integrated, 357 USA) positioned directly on the PCB boards to monitor heater block temperature changes 358 and provide feedback control. The specially designed aluminium heater blocks to hold 200 359 µl PCR tubes and the lid heater to prevent condensation were attached directly on top of the 360 surface mount temperature sensors on the respective PCBs with a heat transferring 361 adhesive (TermoGlue, Termopasty Grzegorz Gasowski, Poland). The flexible resistive 362 heating elements were also attached to the heater blocks. To circumvent the need for 363 specialised docks and eliminate user interpretation of the colorimetric results, a Raspberry 364 Pi Camera (RPi Camera) was used. Eight LEDs (LW T733, Osram, Germany) were 365 assembled on the top side of the lid mount PCB to shine light directly into the reaction tubes 366 to achieve consistent lighting within the device. All the above components were assembled 367 into a 3D printed enclosure (14.3 x 10.8 x 6 cm) specially designed with openings to access 368 to the USB and TCP/IP ports of the RPi. A 20,000 mAh power bank (Anker Power Core, 369 Anker, China) with two 5V, 2A output was used to power the device. A Python based control 370 software was used to control the heating, image the progression of the LAMP assay and 371 store the ‘time-lapse’ images and temperature data within a specified folder. The user can 372 initiate a test by either connecting to a screen via the HDMI port or through simply pairing 373 the device with the mobile app via Bluetooth and selecting the required diagnostic assay. 374 375 3.6. Automated image acquisition and processing through a template matching-based 376 algorithm 377 The LAMP assay (8 separate tubes) was remotely started to initiate heating to 65oC. 378 Images of those test tubes were captured using the inbuilt RPi Camera for every 20 seconds 379 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 15, 2020. ; https://doi.org/10.1101/2020.07.08.20148999doi: medRxiv preprint and were saved in the RPi in the RGB format. Each individual image (90 images in total) 380 consisted of 8 frames for each tube with a black background and these were analysed for 381 30 minutes. As the tube area exposing colour changes was fraction ally small compared to 382 its background, we first extracted each targeted tube frame from the image before applying 383 an image processing algorithm. In order to process these extracted frames, a reference tube 384 was selected as a template, and a template matching algorithm [25] was applied to extract 385 all tubes from the first image. The rationale for the template matching was to search and find 386 the location of a template image in a larger image. It simply slides the template image over 387 the input image to perform the 2 -dimensional convolution and compared values to get the 388 maximum overlap to decide the exact similar areas. Assuming that positions of the test-tube 389 do not change over the time of an experiment, images were cropped in an experiment to 390 obtain the tube frames from the en tire image. These crops are then saved into a 2 -391 dimensional array for RGB colour space (see equation below). Once extracted, RGB format 392 images were converted to YUV format using the following transformation [26] to avoid 393 diffraction and lighting variabilities in different images. 394 ! 𝑌 𝑈 𝑉 % = ! +0.257 + 0.504 + 0.098 −0.148 − 0.291 + 0.439 +0.439 − 0.368 − 0.071 % . ! 𝑅 𝐺 𝐵 % + ! 16 128 128 % 395 In YUV colour space, the Y channel represented the luminance of the colour, while the 396 U and V channels represented the chrominance (Figure 4B). Separating the luminance from 397 the chrominance reduced the effect of light changing and shadow noises in each tested tube 398 [27]. Finally, the chrominance (U, V) channels from the YUV image were considered for 399 image processing. The chrominance (U, V) values of those extracted test tubes were 400 compared with reference orange test tubes in positive control and reference pink image in 401 negative control test tubes to calculate the sum of absolute difference (SAD) for each of the 402 pixel values. After experiments with two images set and fine tuning the threshold values 403 manually, a SAD threshold value was achieved which provides 100% accuracy in the 404 separation of COVID-19 positive and negative samples. 405 406 407 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 15, 2020. ; https://doi.org/10.1101/2020.07.08.20148999doi: medRxiv preprint Figure 4: Fabrication and proce ssing of LAMP data for enhanced detection of 408 SARS-CoV-2. (A) Exterior of a smart diagnostic device ( B) Description of the AI -409 assisted algorithm and image processing. ( C) Pipeline to process images and 410 extraction of colorimetric information. ( D) Schematic o utlining the training of the 411 network for image processing. 412 413 3.7. Artificial intelligence -assisted rap id detection of colour changes associated with the 414 LAMP reaction 415 Deep learning is a subdomain of AI which doesn’t require any domain knowledge to 416 work. However, it learns hidden patterns from examples present in the dataset. A deep 417 learning Convolutional Neural Network (CNN) [28] architecture was proposed with the 418 bespoke 8 layers deep mode as shown in Figure 4C. It consisted of four convolutional layers 419 followed by 2 dense and an output layer. Binary cross-entropy was used as a loss function 420 for optimizing this CNN model. 421 For the training of the network, the dataset with 360 images was shuffled and then split 422 into 9:1 proportion (Figure 4D). 90% of the data was used to train the network and the 423 remaining 10% was exploited to check how the network behaved o n seeing a new image. 424 Training a dataset requires loading numerous images into the memory in a single operation 425 which is an expensive process. Therefore, a data generator was implemented that read the 426 data in batches from the dataset directory and fed it to the model. After multiple experiments, 427 it was observed that the network converged after 6 cycles (epochs) through the dataset. 428 Therefore, we ran an experiment for only 6 epochs to decrease the probability of overfitting. 429 In addition, an additional set of 108 test-tube crops was used to validate the network. The 430 best performing network resulted in an accuracy of 98% in classifying tubes based on their 431 colours (images with better light). 432 433 434 Figure 5: Conventional and AI -assisted interpretation of LAMP results. ( A) 435 Temporal analysis of known positive and negative patient samples from 360 images 436 taken from RPi for visual interpretation of LAMP r esults. ( B) Interpretation of 437 corresponding patient samples by the AI-assisted LAMP results. 438 439 In order to assess the temporal impact of the AI -assisted detection of colour changes 440 (indicative of amplification), the RT -LAMP reaction was run with 3 previously confirmed 441 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 15, 2020. ; https://doi.org/10.1101/2020.07.08.20148999doi: medRxiv preprint positive and negative patient samples in addition to known positive and negative controls. 442 Colour changes were assessed every 5 minutes until the comple tion of the LAMP reaction 443 at 30 minutes. Gradual colour changes were detectable with the naked eye as early as 20 444 minutes post-start of the reaction (Figure 5A). Corresponding samples were run on the newly 445 developed device and temporal and real-time colour changes were monitored as described 446 earlier. Depending upon the viral load in the test sample, a clear colour change was detected 447 and calibrated as early as 20 minutes using device operated processing of the data (Figure 448 5B). Once the positive test control was identified as positive, the test was be stopped, and 449 the results were returned, thus reducing the waiting time and power consumption. 450 As shown in Fig 5B, a template matching algorithm was applied to extract test tubes 451 from images the CNN model was applied to the extracted tubes to train the model. The CNN 452 model was used as a machine learning algorithm to identify colours for each image taken 453 throughout the experiment. Images taken at time tare marked as ‘t’ in the Fig 5B. Once the 454 positive control test and the negative control test gave the correct results consecutively on 455 three occasions, the LAMP based test was be stopped, and the results returned. This 456 approach reduces waiting times for the results and power consumption. 457 458 3.8. Validation of ai-LAMP and comparative performance in clinical settings 459 In order to assess the field application of the optimized assay, we applied the ai-LAMP 460 to purified RNA spiked with miR-cel-miR-39-3p from CoVID-19 patients. A total of 199 swab 461 samples were collected from CoVID-19 clinically suspected patients during routine COVID-462 19 screening at the Royal Lancaster Infirmary (RLI), University Hospitals of Morecambe Bay 463 NHS Foundation Trust UK. The extracted RNA from swab samples were run in parallel for 464 ai-LAMP and two WHO/PHE recommended qRT-PCR targeting the RdRP and N genes of 465 the SARS-CoV-2. This parallel assessment facilitated the assessment of the comparative 466 performance of the ai-LAMP. 467 The RdRP gene-based qRT-PCR detected a total of 67 positives and 132 negatives in 468 a cohort of 199 patients (Figure 6A). In contrast, a higher number of positive (n=88) and 469 lower numbers of negative (n=111) were detected by the qPCR which targeted the N gene 470 (Figure 6B). Interestingly, the ai -LAMP detected a total of 126 positive samples which 471 constituted several times higher than RdRP and N gene -based qPCR, respectively. 472 Comparative analysis of these three molecular detection assays revealed 58 total true 473 positives (TP), 09 false negatives (FN), 64 true negatives (TN), and 68 false positives (FP) 474 in RdRP -based qRT -PCR compared to RdRP -based LAMP (Figure 6A). Similarly, upon 475 comparative analysis of the N gene -based qPCR and RdRP-based LAMP, we observed a 476 total of 74 TP, 14 FN, 59 TN, and 52 FP (Figure 6B). 477 In the current clinical settings, a qRT -PCR targeting two genes (N and RdRP) was 478 conducted to conclusively identify CoVID -19 positive cases and this assay is referred as 479 cumulative (CUM) qRT -PCR. In this scenario, a sample would be considered as positive 480 only if a Ct value of =/> 35 was detected in both N and RdRP-gene based qRT-PCR. Using 481 this approach, we noticed a total of a 70 positive and 129 negative samples and an improved 482 true positive (n=61), false negatives (n=09), true negatives (n=64), and false positives (n=65) 483 limits (Figure 6C). Taken together, the cumulative comparative picture of the qPCR and ai-484 LAMP has identified a superior detection of positive cases (Figure 6D). In order to confirm 485 this detection performance, all ai -LAMP amplification products were visualised by 486 electrophoresis (data not shown), further confirming the aided-detection and improved 487 implication of ai-LAMP in field conditions. 488 Next we determined the detection limit of the ai -LAMP in direct correlation with the 489 standard Ct values of the qPCRs. Plotting of ai-LAMP positive and negative data against the 490 linearity of the Ct values revealed that ai -LAMP carried Cp (cycle number at detection 491 threshold) of up to 37 Ct determined by the RdRP gene -based qPCR (Figure 6E and 492 Supplementary Table 1) or N gene qRT-PCR (Figure 6F and Supplementary Table 1) of the 493 SARS-CoV-2. This detection was approximately 2 Ct values higher than the detection limit 494 of the standard qPCR. Analysis of the first 96 samples, run in parallel for the ai -LAMP, 495 (Figure 6G and Supplementary Table 1) showed a clear demarcation of the positive an d 496 negative samples in the RdRP -gene based ai -LAMP. In order to rule out the quantitative 497 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 15, 2020. ; https://doi.org/10.1101/2020.07.08.20148999doi: medRxiv preprint recovery from spiked miRNA, a qRT-PCR was run on 40 randomly selected RNA samples 498 [23]. Based on the quantitative Ct values for miRNA, all samples showed a marked recovery 499 except a single sample where a low detection of the miRNA was identified (Figure 6H, 500 Supplementary Table 1 and Supplementary Figure 1). 501 502 503 Figure 6. Clinical validation of ai-LAMP. (A-C) Comparative sample positivity between 504 LAMP and RdRP qRT-PCR (A), LAMP and N qRT-PCR (B), LAMP and CUM qRT-PCR 505

(C). (D) The heatmap indicate the relative positive and negative samples among 506 three assays. (D) Linearity chart comparing the LAMP positive/negative samples and 507 their detection based on the RdRP gene-based qRT-PCR. (F) Linearity chart comparing 508 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 15, 2020. ; https://doi.org/10.1101/2020.07.08.20148999doi: medRxiv preprint the LAMP positive /negative samples and their detection based on the N gene -based 509 qRT-PCR. (G) Naked eye detection of the first 96 samples out of the total 199 patients’ 510 samples were processed. ( H) Recovery Ct values of the miRNA from spiked before 511 RNA extraction. 512 Collectively, these data highlight the improved specificity and sensitivity of the AI -513 assisted LAMP assay compar ed to the naked -eye interpretation of the LAMP result, thus 514 enhancing the timely and automated detection, and interpretation of the assay results (see 515 supplementary data). 516 517 4. Discussion 518 The SARS-CoV-2 is now a global pandemic, over 216 countries are currently reporting 519 active infections and the number of daily infections and deaths is continuing to rise, 520 especially in the Americas and South East Asia through a series of multiphasic spread [29]. 521 Currently, there is no licensed vaccine or registered drugs, leaving timely identification of 522 CoVID-19 patients, contact tracing and isolation of positive contacts as the most effective 523 means of containing the pandemic. Among different molecular diagnostic chemistries, LAMP 524 technology provides a promising approach for the rapid and reliable detection in resource -525 limited settings [17]. Recently, the LAMP technology has been widely applied for the 526 identification of West Nile virus, influenza virus, yellow f ever virus, Marburg virus, Ebola 527 virus, Zika virus, and other myriads of viruses [30- 35]. Here, we demonstrated the 528 specificity, sensitivity and utility of a novel ai-LAMP assay for SARS-CoV-2. 529 The genome of SARS -CoV-2 is approximately 30kb in size with a coding capacity of 530 9860 amino acids. All of the β-coronaviruses encodes for structural (replicases, S, E, M and 531 N) genes in the order of 5’ to 3’ in the positive sense genome [5, 36, 37]. A range of qRT -532 PCRs have been proposed and are referred by the Worl d Health Organization [29; 533 https://www.who.int/emergencies/diseases/novel-coronavirus-2019/technical-534 guidance/laboratory-guidance] for diagnosis of SARS-CoV-2. While diagnostic assays can 535 be designed on the most conserved region of the viral genome, most of the routinely applied 536 RT-PCR and RT-LAMP have been targeting the S, N, RdRP, E and ORF1a/b genes mainly 537 due to their high level of transcription and abundance in expression compared to other genes 538 of the SARS-CoV-2 [5, 6, 38]. For the detection of SARS-CoV-2, Chan et al., [23] have 539 targeted and developed a standard RT-LAMP with LoD of 11.2 RNA copies/reaction using 540 invitro RNA transcripts. Yan et al., [16] have adapted the ORF1ab to developed RT-LAMP 541 assay with a detection limit of sensitivities of 2x101 copies per reaction. The majority of these 542 diagnostic assays carry a high level of sensitivity, specificity and repeatability; however, 543 these primarily lack the clinical validation and/or optimization on the synthetic targets. 544 In this study, we have developed and evaluated a novel RT-LAMP in one of the most 545 conserved genes (i.e. RdRP) within the SARS-CoV-2 genome. The RT-LAMP was then 546 directly compared with the currently applied routine diagnostic assays to assess the 547 comparative performance. The RT-LAMP assay developed in this study, could detect as low 548 as 100 copies with an in vitro RNA transcript. Importantly, the RT-LAMP has detected the 549 SARS-CoV-2 RNA in 68/199 (34%) and 52/199 (26%) additional specimens that were tested 550 negative by the RdRP-based qRT-PCR and N-based qRT-PCR, respectively. These 551 findings are interesting, both clinically and epidemiologically due to the high proportion of 552 asymptomatic and mildly symptomatic cases of CoVID-19. These apparently healthy people 553 have been suggested to be a major sources of virus propagation and the basis of epidemics 554 within the community [39- 41]. Therefore, highly sensitive and specific test is needed to 555 identify cases with low viral load. The RT-LAMP detected more positive samples which were 556 otherwise negative by routinely applied qRT-PCR assay. In order to assess the potential 557 false positive identification, we run electrophoreses and sequencing of the N gene. The use 558 of a spiked RNA standard that is not expressed in humans (cel-miR-39-3p) helped to confirm 559 the effective effectiveness of the RNA extraction approach using commercial kits (Qiagen). 560 In addition, we used a fixed total RNA concentration in all experiments allowing for better 561 comparisons across groups. 562 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 15, 2020. ; https://doi.org/10.1101/2020.07.08.20148999doi: medRxiv preprint The main challenges of using the colorimetric approach are background which changes 563 the colour perspective, issues in identifying small changes, bubbles in the test tube s, 564 relatively small area corresponding to colour change and pixel variation due to camera flash 565 and background reflections. The CNN based model has used high -performance computer 566 images to train using these issues and having learned the patterns is able to classify colour, 567 despite the presence of noise. The trained model has successfully moved to Rpi to identify 568 colour changes in test tubes. The study produced 98% accuracy for images taken with better 569 light (Open) and the duration of testing could be dynamically controlled to reduce the length 570 of operating time and heating with a resulting reduction in energy consumption by the device. 571 Despite the SAD based approach resulting in 100% accuracy for the images after 30 572 minutes, this approach failed with other d atasets containing bubbles and different 573

lights as a different threshold value was produced for each image set. 574 Therefore, a convolutional neural network (CNN) approach was utilised in our experiments 575 to generalize the classification for orange and pink test tubes with different background light 576 and bubbles. 577 Collectively, our data showed that the newly established ai -LAMP was highly specific 578 for the detection of SARS -CoV-2 RNA from extracted respiratory tract clinical specimens. 579 The application of this novel LAMP assay may be particularly useful for detecting COVID -580 19 cases with low viral loads and when testing upper respiratory tract specimens (nasal or 581 oral swabs) from patients. D evelopment of ai -LAMP into a multiplex assay which can 582 simultaneously detect other human -pathogenic coronaviruses and respiratory pathogens 583 may further increase its clinical utility in the future. 584 Supplementary Materials: The following are available online at www.mdpi.com/xxx/s1, 585 Figure S1: Optimization and raw data on the LAMP optimization, Table S1: Data on the 586 comparison of the qRT-PCR and LAMP. 587 Author Contributions: Conceptualization, M.M., A.F. and W.B; methodology, M.A.R, W.B. 588 and M.M.; software, A.F.; validation, M.T., N.S.C. and M.A.R.; formal analysis, M.A.R, A.P. 589 M.B; investigation, M.M., M.A.R; resources, M.M., A.F. and W.B; writing —original draft 590 preparation, M.M; writing—review and editing, M.A.R, E.C., I.S., J.V., M.E.K, M.Q.A, M.B., 591 A.P., M.B., M.T., N.S.C., R.S., A.B., P.B., W.H., J.B., J.B., H.W., C.W., M.B., R.L.R., W.B., 592 A.F., and M.M; supervision, R.L.R., A.F., W.B., and M.M.; project administration, M.M; 593 funding acquisition, R.L.R., A.F., W.B., and M.M. All authors have read and agreed to the 594 published version of the manuscript. 595 Funding: The authors wish to express our sincere appreciation to the BBSRC for allowing 596 us to repurpose the LAMP prototypes produced in the grant BB/R012695/1 to be used for 597 SARS-CoV-2 laboratory testing at The University of Lancaster. We would like to thank the 598 support of BBSRC (BB/M008681/1 and BBS/E/I/00001852) and British Council (172710323 599 and 332228521) at Division of Biomedical and Life Sciences, Lancaster University, UK. We 600 would also like to thank Brunel University London and the University of Surrey for providing 601 some financial support to rapidly produce these devices. 602 Acknowledgments: The authors would like to thank the Electronic Technicians William 603 Schkzamian, Gopalakirishnan Jeysundra and Michael Lateo of Brunel University London for 604 their efforts to travel to the University with special permission during the early lockdown 605 period to produce eight laboratory prototypes within 5 days. We thank the Microbiology 606 Department, University Hospitals of Morecambe Bay for access to anonymised patient 607 samples and acknowledge the support of BLS Lancaster University Technicians throughout 608 the lockdown period. We would like to thank Dr Derek Gatherer, Lancaster University, in 609 aligning large SARS-COV-2 genome sequences. 610 Conflicts of Interest: The authors declare no conflict of interest . The funders had no role 611 in the design of the study; in the collection, analyses, or interpretation of data; in the writing 612 of the manuscript, or in the decision to publish the results. 613 614 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 15, 2020. ; https://doi.org/10.1101/2020.07.08.20148999doi: medRxiv preprint

615 1. Zhu, N.; Zhang, D.; Wang, W.; Li, X.; Yang, B.; Song, J.; Zhao, X.; Huang, B.; Shi, W.; 616 Lu, R.; Niu, P.; Zhan, F.; Ma, X.; Wang, D.; Xu, W.; Wu, G.; Gao, G.F.; Tan, W.; China 617 Novel Coronavirus Investigating and Research Team. A novel coronavirus from patients 618 with pneumonia in China, 2019. N. Engl. J. Med. 2020, 382(8), 727-733. 619 2. Chan, J. F.; Yuan, S.; Kok, K.H.; To, K.K.; Chu, H.; Yang, J.; Xing, F.; Liu, J.; Yip, C.C.; 620 Poon, R.W.; Tsoi, H.;W.; Lo, S.K.; Chan, K.H.; Poon, V.K.; Chan, W.M.; Ip, J.D.; Cai, 621 J.P.; Cheng, V.C.; Chen, H.; Hui, C.K.; Yuen, K.Y. A familial cl uster of pneumonia 622 associated with the 2019 novel coronavirus indicating person-to-person transmission: a 623 study of a family cluster. Lancet. 2020a, 395, 514 –523. 624 3. Zhou, P.; Yang, X.L.; Wang, X.G.; Hu, B.; Zhang, L.; Zhang, W.; Si, H.R.; Zhu, Y.; Li, 625 B.; Huang, C.L.; Chen, H.D.; Chen, J.; Luo, Y.; Guo, H.; Jiang, R.D.; Liu, M.Q.; Chen, 626 Y.; Shen, X.R.; Wang, X.; Zheng, X.S.; Shi, Z.L. A pneumonia outbreak associated with 627 a new coronavirus of probable bat origin. Nature. 2020, 579(7798), 270-273. 628 4. Decaro, N.; Lorusso, A. Novel human coronavirus (SARS-CoV-2): A lesson from animal 629 coronaviruses. Vet Microbiol. 2020, 244, 108693. 630 5. Chan, J.F.; Kok, K.H.; Zhu, Z.; Chu, H.; To, K.K.; Yuan, S.; Yuen, K.Y. Genomic 631 characterization of the 2019 novel human-pathogenic coronavirus isolated from a 632 patient with atypical pneumonia after visiting Wuhan. Emerg. Microbes Infect. 2020, 633 9(1), 221–236. 634 6. Kim, D.; Lee, J.; Yang, J.; Kim, J.W.; Kim, V.N.; Chang, H. The Architecture of SARS-635 CoV-2 Transcriptome. Cell. 2020, 181(4):914-921.e10. 636 7. Peiris, J.S.; Lai, S.T.; Poon, L.L.; Guan, Y.; Yam, L.Y.; Lim, W.; Nicholls, J.; Yee, W.K.; 637 Yan, W.W.; Cheung, M.T.; Cheng, V.C.; Chan, K.H.; Tsang, D.N.; Yung, R.W.; Ng, T.K.; 638 Yuen, K.Y .; SARS study group. Coronavirus as a possible cause of severe acute 639 respiratory syndrome. Lancet. 2003, 361(9366), 1319-25. 640 8. Woo, P.C.; Yuen, K.Y. Severe acute respiratory syndrome coronavirus as an agent of 641 emerging and reemerging infection. Clin. Microbiol. Rev. 2007, 20(4), 660–694. 642 9. Yuen, K.Y. Middle East respiratory syndrome coronavirus: another zoonotic 643 betacoronavirus causing SARS-like disease. Clin. Microbiol. Rev. 2015, 28(2), 465-522. 644 10. Javaid, A.; Hussain, N. Mutational analysis of Forkhead box P3 gene in Pakistani 645 Human Immunodeficient Virus Patients. Pakistan J. Zool. 2019, 51(5), 1987-1990 646 11. Afshan, G.; Zulfiqar, S.; Mehboob, S.; Khan, M.T.J.; Shakoori, A. Production of 647 Antibodies against Hepatitis C Virus Envelope Glycoprotein E2- A Potential Vaccine 648 Against HCV Infection. Pakistan J. Zool. 2019, 51(6), 2311-2322 649 12. Shahid, M.; Amin, I.; Afzal, S.; Fatima, Z.; Idrees, M. Comparative Analysis of 650 Immunological and Genomic Outcomes of Dengue Virus Outbreak in Pakistan. Pakistan 651 J. Zool. 2019, 51(05), 1971-1974 652 13. WHO, Novel Coronavirus – China 2020a. https://www.who.int/csr/don/12-january-2020-653 novel-coronavirus-china/en/ 654 14. Vogels, C.B.F.; Brito, A.F.; Wyllie, A.L.; Fauver, J.R.; Ott, I.M.;Kalinich, C.C.; 655 Petrone, M.E.; Casanovas-Massana, A.; Muenker, M.C.; Moore, A.J.; Klein, J.; Lu, P.; 656 Lu-Culligan, A.; Jiang, X.; Kim, D.J.; Kudo, E.; Mao, T.; Moriyama, M.; Ji Oh, J.E.; 657 Park, A.; Silva, J.; Song, E.; Takehashi, T.; Taura, M.; Tokuyama, M.; 658 Venkataraman, A.; El Weizman, O.; Wong, P.; Yang, Y.; Cheemarla, N.R.; White, E.; 659 Lapidus, S.; Earnest, R.; Geng, B.; Vijayakumar, P.; Odio, C.; Fournier, J.; Bermejo, S.; 660 Farhadian, S.; Cruz, C.D.; Iwasaki, A.; Ko, A.I.; Landry, M.; Foxman, E.F.; Grubaugh, 661 N.D. Analytical sensitivity and efficiency comparisons of SARS-COV-2 qRT-PCR 662 assays. medRxiv. 2020, 2020.03.30.20048108. Available 663 from: https://medrxiv.org/content/early/2020/04/01/2020.03.30.20048108. 664 15. Kashir, J.; Yaqinuddin, A. Loop mediated isothermal amplification (LAMP) assays as a 665 rapid diagnostic for COVID-19. Medical Hypotheses. 2020, 141. 666 doi:10.1016/j.mehy.2020.109786. 667 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 15, 2020. ; https://doi.org/10.1101/2020.07.08.20148999doi: medRxiv preprint 16. Yan, C.; Cui, J.; Huang, L.; Du, B.; Chen, L.; Xue, G.; Li, S.; Zhang, W.; Zhao, L.; Sun, 668 Y.; Yao, H.; Li, N.; Zhao, H.; Feng, Y.; Liu, S.; Zhang, Q.; Liu, D.; Yuan, J. Rapid and 669 visual detection of 2019 novel coronavirus (SARS-CoV-2) by a reverse transcription 670 loop-mediated isothermal amplification assay. Clin. Microbiol. Infec. 2020, 26(6), 773-671 779. 672 17. Mori, Y.; Notomi, T. Loop-mediated isothermal amplification (LAMP): Expansion of its 673 practical application as a tool to achieve universal health coverage. J. Infect. 674 Chemother. 2020, 26(1), 13-17. 675 18. Mahony, J.; Chong, S.; Bulir, D.; Ruyter, A.; Mwawasi, K.; Waltho, D. Development of a 676 sensitive loop-mediated isothermal amplification assay that provides specimen-to-result 677 diagnosis of respiratory syncytial virus infection in 30 minutes. J. Clin. Microbiol. 2013, 678 51(8), 2696-701. 679 19. Ganguli, A.; Ornob, A.; Yu, H.; Damhorst, G.L.; Chen, W.; Sun, F.; Bhuiya, A.; 680 Cunningham, B.T.; Bashir, R. Hands-free smartphone -based diagnostics for 681 simultaneous detection of Zika, Chikungunya, and Dengue at point-of-care. Biomed. 682 Microdevices. 2017, 19(4), 73. 683 20. Kaarj, K.; Akarapipad, P.; Yoon, J.Y. Simpler, Faster, and Sensitive Zika Virus Assay 684 Using Smartphone Detection of Loop-mediated Isothermal Amplification on Paper 685 Microfluidic Chips. Sci Rep. 2018, 8(1): 1-11. 686 21. Yu, L.; Wu, S.; Hao, X.; Dong, X.; Mao, L.; Pelechano, V.; Chen, W.H.; Yin, X. Rapid 687 Detection of COVID -19 Coronavirus Using a Reverse Transcriptional Loop -Mediated 688 Isothermal Amplification (RT -LAMP) Diagnostic Platform. Clin. Chem. 2020, 66(7), 689 975–977. 690 22. Nguyen T, Bang DD, Wolff A. 2019 Novel coronavirus disease (COVID-19): Paving the 691 road for rapid detection and point-of-care diagnostics. Micromachines (Basel). 2020 692 March;11(3):306. 693 23. Chan, J.F.; Yip, C.C.; To, K.K.; Tang, T.H.; Wong, S.C.; Leung, K.H.; Fung , A.Y.; Ng, 694 A.C.; Zou, Z.; Tsoi, H.W.; Choi, G.K.; Tam, A.R.; Cheng, V.C.; Chan, K.H.; Tsang, O.T.; 695 Yuen, K.Y. Improved molecular diagnosis of COVID-19 by the novel, highly sensitive 696 and specific COVID-19-RdRp/Hel real-time reverse transcription-PCR assay validated 697 in vitro and with clinical specimens. J. Clin. Microbiol. 2020, 58(5), e00310-20. 698 24. Katoh, K.; RozewickI, J .; Yamada, K.D. MAFFT Online Service: Multiple Sequence 699 Alignment, Interactive Sequence Choice and Visualization. Brief Bioinform. 2019, 20(4), 700 1160-1166. 701 25. Sun, Y.; Mao, X.; Hong, S.; Xu, W.; Gui, G. Template Matching -Based Method for 702 Intelligent Invoice Information Identification. IEEE Access. 2019, 7, 28392-28401. 703 26. Yang, G.; Li, H.; Zhang, L.; Cao, Y. Research on a skin color detection algorithm based 704 on self adaptive skin color model; 2010 International Conference on Communications 705 and Intelligence Information Security, Nanning, China. IEEE Xplore. 2010, 266-270. 706 27. Al-Tairi, Z.H.; Rahmat, R.W.O.; Saripan, M.I.; Sulaiman, P.S. Skin Segmentation Using 707 YUV and RGB Color Spaces. J. Inf. Process Syst. 2014, (10)2, 283-299. 708 28. Sharif M, Khan MA, Rashid M.; Yasmin, M.; Afza, F.; Tanik, U.J. Deep CNN and 709 geometric features-based gastrointestinal tract diseases detection and classification 710 from wireless capsule endoscopy images. J. Ex. Theor. Artif. In. 2019, 711 doi.org/10.1080/0952813X.2019.1572657 712 29. WHO, Coronavirus disease (COVID-19) pandemic 2020b. 713 https://www.who.int/emergencies/diseases/novel-coronavirus-2019 714 30. Kurosaki, Y.; Grolla, A.; Fukuma, A.; Feldmann, H.; Yasuda, J. Development and 715 evaluation of a simple assay for Marburg virus detection using a reverse transcription-716 loop-mediated isothermal amplification method. J. Clin. Microbiol. 2010, 48, 2330–2336. 717 31. Ge, Y.; Wu, B.; Qi, X.; Zhao, K.; Guo, X.; Zhu, Y.; Qi, Y.; Shi, Z.; Zhou, M.; Wang, H.; 718 Cui, L. Rapid and sensitive detection of novel avian-origin influenza A (H7N9) virus by 719 reverse transcription loop-mediated isothermal amplification combined with a lateral- 720 flow device. PLoS One. 2013, 8(8), e69941. 721 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 15, 2020. ; https://doi.org/10.1101/2020.07.08.20148999doi: medRxiv preprint 32. Kwallah, A.; Inoue, S.; Muigai, A.W.; Kubo, T.; Sang, R.; Morita, K.; Mwau, M. A real-722 time reverse transcription loop-mediated isothermal amplification assay for the rapid 723 detection of yellow fever virus. J. Virol. Methods. 2013, 193(1), 23-7. 724 33. Cao, Z.; Wang, H.; Wang, L.; Li, L.; Jin, H.; Xu, C.; Feng, N.; Wang, J.; Li, Q.; Zhao, Y.; 725 Wang, T.; Gao, Y.; Lu, Y.; Yang, S.; Xia, X. Visual detection of west nile virus using 726 reverse transcription loop-mediated isothermal amplification combined with a vertical 727 flow visualization strip. Front. Microbiol. 2016, 7, 554. 728 34. Xu, C.; Wang, H.; Jin, H.; Feng, N.; Zheng, X.; Cao, Z.; Li, L.; Wang, J.; Yan, F.; Wang, 729 L.; Chi, H.; Gai, W.; Wang, C.; Zhao, Y.; Feng, Y.; Wang, T.; Gao, Y.; Lu, Y.; Yang, S.; 730 Xia, X. Visual detection of Ebola virus using reverse transcription loop -mediated 731 isothermal amplification combined with nucleic acid strip detection. Arch. Virol. 2016, 732 161(5), 1125-33. 733 35. Chotiwan, N.; Brewster, C.D.; Magalhaes, T.; Weger -Lucarelli, J.; Duggal, N.K. ; 734 Rückert, C.; Nguyen, C.; Garcia Luna, S.M.; Fauver, J.R.; Andre, B.; Gray, M.; Black, 735 W.C.; Kading, R.C.; Ebel, G.D.; Kuan, G.; Balmaseda, A.; Jaenisch, T.; Marques, 736 E.T.A.; Brault, A.C.; Harris, E.; Foy, B.D.; Quackenbush, S.L.; Perera, R.; Rovnak, J. 737 Rapid and specific detection of Asian- and African- lineage Zika viruses. Sci. Transl. 738 Med. 2017, 9(388), eaag0538. 739 36. Chen, L.; Liu, W.; Zhang, Q.; Xu, K.; Ye, G.; Wu, W.; Sun, Z.; Liu, F.; Wu, K.; Zhong, 740 B.; Mei, Y.; Zhang, W.; Chen, Y.; Li, Y.; Shi, M.; Lan, K.; Liu, Y. RNA based mNGS 741 approach identifies a novel human coronavirus from two individual pneumonia cases in 742 2019 Wuhan outbreak. Emerg. Microbes Infect. 2020, 9(1), 313–319. 743 37. Lu, R., Zhao, X., Li, J., Niu, P., Yang, B., Wu, H., Wang, W., Song, H., Huang, B., Zhu, 744 N., Bi, Y., Ma, X., Zhan, F., Wang, L., Hu, T., Zhou, H., Hu, Z., Zhou, W., Zhao, L., Chen, 745 J., Meng, Y.; Wang, J.; Lin, Y.; Yuan, J.; Xie, Z.; Ma, J.; Liu, W.J.; Wang, D.; Xu, W.; 746 Holmes, E.C.; Gao, G.F.; Wu, G.; Chen, W.; Shi, W.; Tan, W. Genomic characterisation 747 and epidemiology of 2019 novel coronavirus: implications for virus origins and receptor 748 binding. The Lancet. 2020, 395, 565–574. 749 38. Yang, W.; Dang, X.; Wang, Q.; Xu, M.; Zhao, Q.; Zhou, Y.; Zhao, H.; Wang, L.; Xu, Y.; 750 Wang, J.; Han, S.; Wang, M.; Pei, F.; Wang, Y. Rapid Detection of SARS-CoV-2 Using 751 Reverse transcription RT-LAMP method. medrxiv, 2020, 752 https://doi.org/10.1101/2020.03.02.20030130 753 39. Wei, M.; Yuan, J.; Liu, Y.; Fu, T.; Yu, X.; Zhang, Z.J. Novel coronavirus infection in 754 hospitalized infants under 1 year of age in China. JAMA. 2020, 323(13), 1313-1314. 755 40. Shahid, M.; Amin, I.; Afzal, S.; Fatima, Z.; Idrees, M. Comparative Analysis of 756 Immunological and Genomic Outcomes of Dengue Virus Outbreak in Pakistan. Pakistan 757 J. Zool. 2019, 51(5), 1971-1974 758 41. Sodi, R.; Eastwood, J.; Caslake, M.; Packard, C.J.; Denby, L. Relationship Between 759 Circulating microRNA-30c With Total- And LDL-cholesterol, Their Circulatory 760 Transportation and Effect of Statins. Clin Chim Acta. 2017, 466, 13-19 761 © 2020 by the authors. Submitted for possible open access publication under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). 762 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 15, 2020. ; https://doi.org/10.1101/2020.07.08.20148999doi: medRxiv preprint