key: cord-0891123-gdblhjyw
authors: Xie, Shengsong; Tao, Dagang; Fu, Yuhua; Xu, Bingrong; Tang, You; Steinaa, Lucilla; Hemmink, Johanneke D; Pan, Wenya; Huang, Xin; Nie, Xiongwei; Zhao, Changzhi; Ruan, Jinxue; Zhang, Yi; Han, Jianlin; Fu, Liangliang; Ma, Yunlong; Li, Xinyun; Liu, Xiaolei; Zhao, Shuhong
title: Rapid visual CRISPR assay: a naked-eye colorimetric detection method for nucleic acids based on CRISPR/Cas12a and convolutional neural network
date: 2021-07-18
journal: bioRxiv
DOI: 10.1101/2021.07.17.452802
sha: 972a7ba72c3cf747abf15c94183bd4d324ada7d7
doc_id: 891123
cord_uid: gdblhjyw

Rapid diagnosis based on naked-eye colorimetric detection remains challenging, but it could build new capacities for molecular point-of-care testing (POCT). In this study, we evaluated the performance of 16 types of single-stranded DNA-fluorophore-quencher (ssDNA-FQ) reporters for use with CRISPR/Cas12a based visual colorimetric assays. Among them, 9 ssDNA-FQ reporters were found to be suitable for direct visual colorimetric detection, with especially very strong performance using ROX-labeled reporters. We optimized the reaction concentrations of these ssDNA-FQ reporters for naked-eye read-out of assay results (no transducing component required for visualization). Subsequently, we developed a convolutional neural network algorithm standardize and to automate the analytical colorimetric assessment of images and integrated this into the MagicEye mobile phone software. A field-deployable assay platform named RApid VIsual CRISPR (RAVI-CRISPR) based on a ROX-labeled reporter with isothermal amplification and CRISPR/Cas12a targeting was established. We deployed RAVI-CRISPR in a single tube towards an instrument-less colorimetric POCT format that requires only a portable rechargeable hand warmer for incubation. The RAVI-CRISPR was successfully used for the single-copy detection of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and African swine fever virus (ASFV). Our study demonstrates this novel RAVI-CRISPR system for distinguishing different pathogenic nucleic acid targets with high specificity and sensitivity as the simplest-to-date platform for rapid pen-side testing.

Nucleic acid-based assays have been widely adopted in many fields, for example, in the diagnosis of infectious diseases 1 , genotyping 2 , and food safety testing 3 . Infectious diseases are presently the focus of considerable research attention to resolve longstanding public health problems from "classical" diseases to newly emerging epidemic outbreaks. In particular, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the novel RNA virus responsible for the COVID-19 pandemic, has a major impact on human health worldwide 4 , infecting over 178 million people as of June 2021, reported to the World Health Organization (WHO) (https://covid19.who.int). In order to control COVID-19 pandemic, there is an urgent need to test for the SARS-CoV-2 in as many people as possible 5, 6, 7 . Similarly, animal production is periodically subject to widespread viral epidemics, as in the case of African swine fever virus (ASFV), a DNA virus that infects domestic pigs and wild boars with a mortality rate of up to 100% 8 . Although agricultural epidemics receive less attention than human epidemics, the early diagnosis of ASFV is necessary to monitor circulating virus for the control and prevention of current and future outbreaks 9,10 , which have major impacts on food production and economy 11 .

Although current PCR-based clinical diagnostic methods for the detection of viral nucleic acids are highly sensitive, these methods require expensive equipment, trained technicians, and clean laboratory environments, thus limiting their application in developing nations and rural areas without clinical infrastructure 12 .

Thus, increasing the affordability of point of care testing (POCT) 13 , and moving diagnostic testing for viral pathogens such as COVID-19 or ASF from the laboratory to pen-side detection could be potentially transformative for epidemiological efforts.

Although real-time quantitative PCR (qPCR)-based nucleic acid detection is the current gold standard for SARS-CoV-2 and ASFV diagnosis 14, 15 , isothermal amplification techniques, such as Loop-Mediated Isothermal Amplification (LAMP) and Recombinase Polymerase Amplification (RPA), represent comparatively easy, accessible, and reliable alternatives 16, 17 . In particular, these methods can combine a variety of portable read-out types for POCT, such as paper lateral flow test strips or fluorescence and colorimetric 4 assays 18, 19, 20 . LAMP/RPA-based detection methods have been rapidly developed and deployed for both SARS-CoV-2 or ASFV owing to their high efficiency of amplification 21, 22, 23, 24 . However, the vast majority of these LAMP/RPA-based assays depend on non-specific nucleic acid detection, meaning that unintended LAMP/RPA amplicons cannot be distinguished from the amplicons of correct targets, leading to unreliable test results 25, 26 .

To overcome this issue, nucleic acid detection methods based on combination of Clusters of Regularly Spaced Short Palindrome Repeats (CRISPR) and LAMP or RPA were developed, including SHERLOCK, DETECTR, HOLMES, and CDetection 27, 28, 29, 30 .

These diagnostic strategies use Cas proteins, which have collateral nuclease activity to specifically target double-strand DNA (dsDNA) or single-strand RNA (ssRNA) and a protospacer adjacent motif (PAM) sequence, thereby effectively avoiding false positive results in isothermal amplification 31, 32 .

Moreover, some studies have reported that Cas-based detection methods can distinguish differences at the single nucleotide level by integrating an optimized CRISPR RNA (crRNA) 29, 30, 33 , and several improvements to CRISPR-based detection have been developed for POCT applications, such as CRISPR-based lateral flow strip assay 34, 35 , CRISPR-based fluorescent cleavage assay 36, 37 , plasmonic CRISPR/Cas12a assay 38, 39 , and Cas12a-modulated fluorescence resonance energy transfer assay that incorporates nanomaterials for nucleic acid sensing 40 . In addition, mobile phone applications (apps) and mobile phone-based devices have been developed to support the evaluation of lateral flow strips or fluorescence signal read-outs 41, 42, 43 . Other studies have established methods that seek to simplify the reaction process like the All-In-One Dual CRISPR/Cas12a assay 44 and a CRISPR/Cas12a based portable biosensor 45 .

In this study, to establish an accurate and convenient method based on colorimetric detection of viral nucleic acids, we developed a practical POCT detection system that does not require advanced training or specialized instruments. We demonstrated the proof-of-concept in RApid VIsual CRISPR (RAVI-CRISPR) system for sensitive testing of simulated SARS-CoV-2 and real ASFV infection samples by a naked-eye detection. 5 Subsequently, we developed a convolutional neural network algorithm to standardize and automate the analytical colorimetric assessment of images and integrated this into a MagicEye mobile phone software for rapid detection of SARS-CoV-2 and ASFV.

To identify the best reporter for the RAVI-CRISPR assay, we screened a total of 16 types of ssDNA-FQ reporters using an exonuclease I digestion and subsequently optimized their concentrations for direct application in RAVI-CRISPR assays (Supplementary Table   S1 ). The two main steps in the assay consisted of an exonuclease I digestion and a reporter signal read-out (Fig. 1A ). The colors emitted by different types of degraded ssDNA-FQ reporters in reaction solution were then compared under a blue or UV light transilluminator, as well as with direct assessment by the naked eye using no excitation light. In addition, the fluorescence signal intensities of the intact or degraded ssDNA-FQ reporters were measured using a fluorescence microplate reader to establish background signals. The results showed that 14 of the degraded ssDNA-FQ reporters can emit strong fluorescence signals under the UV light transilluminator (Fig. 1B) .

Interestingly, we observed by the naked eye that more than 9 of the degraded ssDNA-FQ reporters changed the color of the reaction solution in the PCR test tube without excitation light (Fig. 1B, Supplementary Table S1 ). For instance, the degraded ROX-or Texas RED-labeled reporter reactions exhibited obvious color changes from blue to red ( Fig. 1B) . However, the degraded Cy5-and Cy5.5-labeled reporters were only found to emit fluorescence signals under UV light, while the degraded Cy7-labeled reporter did not emit any fluorescent signal nor change the color in reaction solutions (Fig. 1B) . In conclusion, these results indicated that some ssDNA-FQ reporter modifications had better optical properties than others.

In the naked eye detection of nucleic acids, the optimal concentrations of ssDNA-FQ reporters significantly improved the visual colorimetry ( Supplementary Fig. S1A-1P ). 6 However, for enhanced fluorescence visualization assays, background noise was concurrently increased with increasing signals for VIC, TET, and FAM-labeled reporters, resulting in potential interference with detection accuracy (Supplementary Fig. S1A-1P ).

We further confirmed, using a fluorescence microplate reader, that for some VIC, TET or FAM-labeled ssDNA-FQ reporters, higher concentrations resulted in excessive background fluorescence signals (Fig. 1C ). In conclusion, an optimal concentration should be determined and adjusted for each ssDNA-FQ reporter type depending on the method used to assess the read-out by either fluorescence excitation or the naked eye evaluation with no excitation light. For instance, for a more accurate colorimetric naked eye detection, the optimal concentration was the ROX-labeled reporter at ≥ 5 μM 

In order to develop a highly sensitive diagnostics for the nucleic acids of SARS-CoV-2 using the RAVI-CRISPR system and the naked eye detection, five crRNAs for the SARS-CoV-2 E gene (E-crRNA-1 to 5) and nine crRNAs for the N gene of SARS-CoV-2 (N-crRNA-1 to 9) were screened for their efficiency of target cleavage ( Supplementary Fig.   S2A ). First, we compared the activities of these crRNAs using PCR products and corresponding ssDNA activators as templates. The results showed that E-crRNA-2 and E-crRNA-5 exhibited the highest activity for detection of the E gene ( Supplementary Fig.   S2B ), while N-crRNA-5, N-crRNA-6, N-crRNA-7, and N-crRNA-9 showed sufficient activity for detection of the SARS-CoV-2 N gene ( Supplementary Fig. S2C ). Subsequent experiments used only E-crRNA-5 and N-crRNA-9 for the detection of SARS-CoV-2.

To further evaluate the sensitivity of the RAVI-CRISPR technology for SARS-CoV-2 detection, in vitro transcribed RNAs of the N and E genes from two synthetic plasmids, 7 each carrying one of the target genes, were used as a template ( Fig. 2A) . Following a concentration gradient, we determined that the limit of detection (LoD) for amplification by the RT-LAMP assay was 40 copies/μL (Fig. 2B ). Subsequently, the amplified product of RT-LAMP N gene was digested by Cas12a and subjected to the detection. The result

showed that the sensitivity of this technique reached 40 copies/μL (Fig. 2C ). The fluorescence signal intensity of the RAVI-CRISPR assay was further measured using a fluorescence microplate reader (Fig. 2D) , which confirmed the detection by the naked eye with or without excitation light. In addition, we tested both a lateral-flow paper strip assay integrated with CRISPR/Cas12a ( Table S2 ) showed comparable sensitivity for the detection of SARS-CoV-2 E gene with that of the RAVI-CRISPR assay. Specifically, the Ct values observed in the RT-qPCR assay at the sample concentration ≤ 5.8 copies/μL were ≥ 40 ( Supplementary Fig. S3E ), indicating a negative result. These results 8 confirmed the sensitivity of this technique to be 58 copies/μL for the SARS-CoV-2 E gene. Finally, we found that the RAVI-CRISPR system has high specificity for detecting the SARS-CoV-2 E gene and it can clearly distinguish the targeted E gene from different SARS family viruses ( Supplementary Fig. S3F ). This result was further confirmed with the CRISPR/Cas12a -lateral flow paper strip combination assay ( Supplementary Fig.   S3G ). Thus, the RAVI-CRISPR systems using RT-LAMP coupled with CRISPR/Cas12a allowed a robust detection of the SARS-CoV-2 E and N genes with high sensitivity and specificity.

Considering our result showing effective detection of SARS-CoV-2, we next developed a fast, sensitive, and reliable RAVI-CRISPR-based assay for ASFV using a highly active crRNA identified in our previous study 37 . The strategy for ASFV detection was similar to that for SARS-CoV-2 except that a standard LAMP for DNA amplification was used rather than RT-LAMP since ASFV is a DNA virus (Fig. 3A) . We first evaluated the sensitivity of ASFV p72 by RAVI-CRISPR visual detection using a concentration gradient to determine the LoD to be 7 copies/μL (Fig. 3B ). Subsequently, we assessed our Table S3 ).

Moreover, the read-outs of the RAVI-CRISPR assay were further validated by a multifunctional microplate reader (Fig. 3E ) and the CRISPR/Cas12a-integrated lateralflow paper strip assay (Fig. 3F ). All these results indicated that the detection of ASFV p72 using the CRISPR/Cas12a-based assay can be accurately applied in the clinical screening of ASF samples (Fig. 3G ). In addition, we evaluated the optimal reaction time for RAVI-CRISPR read-out using a p72-containing plasmid as template. The result 9 showed that the shortest reaction time necessary for the LAMP reaction to reach saturation was around 30 min, while the minimum reaction time required for Cas12a cleavage in the naked eye detection was around 5 min (Fig. 3H) . Therefore, this optimized RAVI-CRISPR assay including only 35 min of incubation can detect as few as 7 copies/μL of virus particles per reaction, which was further verified by the screening of ASF infection samples (Fig. 3H) .

In order to provide the RAVI-CRISPR assay as a convenient package including a standardized evaluation mode, we also developed the MagicEye mobile phone app that employed two machine learning models based on convolutional neural networks to automatically interpret the results of nucleic acids detection in multiple reaction tubes (Fig. 4 ). The first model was for object detection, which used a Single Shot MultiBox Detector (SSD) algorithm to identify the images of reaction tubes 46 . The second model was for binary classification of either positive or negative signals to resolve the operatorassociated inconsistencies across the naked eye detection.

To train the first model to recognize the images of PCR tubes, we used 1554 single-PCR tube images, 405 multiple PCR tube images, 283 8-tube PCR strip images, and 301 mixed images (8-tube PCR strip mixed with single-PCR tube images) which were manually labeled (Supplementary Table S4 ). Testing of the object detection model showed that this algorithm could identify PCR tubes in all cases ( Supplementary Fig. S4) with the detection rates at 100% for both single and multiple PCR tube images and > 99% for 8-tube PCR strip images (Supplementary Table S5 ). Subsequently, the binary classification model was trained to distinguish colorimetric changes in the tubes containing the RAVI-CRISPR reactions with either positive or negative detection signals (Supplementary Table S6 ). We found that the F1-Measure, which was used to evaluate the accuracy of predictions in two-class (binary) classification problems, could be increased to 100% by using an optimal ROX-labeled reporter in concentration > 1 μM ( Supplementary Fig. S5 ). This result meant that the model performed well on images captured with no excitation light when the probe concentration was relatively high ( Supplementary Fig. S6 ).

In order to evaluate the robustness of the model at critical concentrations of DNA samples, using images captured under no excitation light as an example, we tested the model on samples with probe concentrations of 1 μM and DNA concentrations ranging from 0 to 7 × 10 5 copies/μL. The result showed that even at very low concentration of DNA template, the model could discriminate negative and positive signals with 100% accuracy (Supplementary Fig. S7, Table S7 ). We also found that the sensitivity of the image analysis app was much higher than that of the naked eye read-out, indicating the power of the app to be used in circumstances where the naked eye evaluation is the only option.

To further evaluate the classification model, we examined the ConvNet outputs at the intermediate training stage of the convolutional model and found that many activation channels were focused on the tip of the PCR tube ( Supplementary Fig. S8) , suggesting that the model may classify results primarily by distinguishing tip color where the signal was the strongest, in the same way as the naked eye observation. Finally, we deployed this algorithm on a cloud server as the MagicEye mobile application for accurate and rapid analysis of nucleic acid detection signals.

The two-step RAVI-CRISPR assay requires opening of the tube after RT-LAMP or LAMP reaction, which may generate aerosols, potentially causing false-positive results.

To reduce the likelihood of contamination and improve the convenience of visualization process for POCT of COVID-19 or ASF, we next established a single-tube RAVI-CRISPR assay. The underlying principle and operation of the assay relied on an isothermal amplification with CRISPR/Cas12a-based detection (Fig. 5A ). To avoid crosscontamination, mineral oil was added to cover the extracted nucleic acids and isothermal amplification solution. Cas12a and crRNA ribonucleoproteins (RNPs) reagents were preadded inside of the tube lid. Following the isothermal amplification, RNPs reagents were mixed with amplification reagents by hand shaking (Fig. 5A) . The whole process did not require the lid to be re-opened, thus avoiding the possibility of aerosol contamination.

We next tested the combined single-tube RAVI-CRISPR/MagicEye system with simulated in vitro transcripts or real clinical samples and found that the SARS-CoV-2 E and N genes and the ASFV p72 genes were all successfully detected (Fig. 5B) . To facilitate the pen-side, instrument-less detection of virus, we used a portable rechargeable hand warmer to provide temperatures from 35 to 60 ℃ for the amplification incubation.

To enhance the fluorescence signal for the pen-side detection, a small flashlight capable of UV and blue light emission were approved to be sufficient (Fig. 5C ). Analysis by a smartphone with the MagicEye app successfully replicated the result of detection by the naked eye for these samples (Supplementary Fig. S9, Table S8 ), thereby demonstrating this single-tube RAVI-CRISPR assay instrumented only with portable rechargeable hand warmer and MagicEye app to be readily deployed in the pen-side POCT of for SARS-CoV-2 or ASFV among other viral pathogens.

pandemic caused by SARS-CoV-2, pose overwhelming threats to human health and global economic stability. Thus, the development of affordable and convenient POCT diagnostic tools is both warranted and urgent. In the present study, we have developed a rapid, single-tube, instrument-less, colorimetric POCT assay ("RAVI-CRISPR") for the detection of nucleic acids as a demonstration of the proof-of-concept for its sensitive detection of diverse pathogenic nucleic acids by the naked eye. We also developed the MagicEye mobile app for standardized diagnostic image analysis of color changes in the reaction tubes. The main feature of this POCT assay is that ssDNA-FQ reporters are used as colorimetric sensors to produce changes in the color of the reaction mixture dependent on the concentrations of the reporters and specifically targeted nucleic acids that are cleaved by Cas12a. To our knowledge, this is the simplest CRISPR based-nucleic acid platform reported to date, since no transducing component is required for visualization (Supplementary Table S9) .

Although previous studies have reported that LAMP reaction can amplify targeted nucleic acids with high efficiency, this reaction produces long concatemeric amplicons, which make sequence-specific detection of LAMP products analytically challenging 26 . A recent study reported that the use of CRISPR/Cas12a in combination with a sequencespecific plasmonic LAMP assay can provide orthogonal color read-outs 47 . However, plasmonic gold nanoparticles (AuNPs) are difficult to produce and their waste is hard to manage, thus limiting their application. To resolve these issues, ssDNA-FQ reporters are proved to be straightforward to synthesize and environmentally harmless compared to AuNPs. We found that the ssDNA-FQ reporter concentration exerts a strong impact on the direct visual detection of nucleic acids. By adjusting the concentration of ROXlabeled reporter, a clear naked-eye observation of the detected nucleic acids became feasible without excitation light. In comparison, colorimetric detection by our RAVI-CRISPR system using ROX-labeled reporter produced stronger color changes than that of the AIOD-CRISPR assay, which used FAM-labeled reporters 44 Table S9 ). Our MagicEye can also improve the detection sensitivity of RAVI-CRISPR assays. The accurate diagnosis of ASFV infection in pigs by MagicEye and visual inspection confirmed the reliability of the RAVI-CRISPR assay. This assay was successfully used for the single-copy detection of ASFV samples in a single-tube incubated only by a portable rechargeable hand warmer. Moreover, this system does not require an electrical power source, therefore it can be feasibly deployed for pen-side testing in rural areas with limited resources.

The partial fragment 780 bp of ASFV (GenBank accession number: MK 333180) p72 gene was synthesized (TSINGKE, Beijing, China) and cloned into the pMD18T vector 

Five crRNAs targeting SARS-CoV-2 E and nine crRNAs targeting SARS-CoV-2 N genes were designed by CRISPR-offinder software 48 (Supplementary Table S10 ) and

complemented to the target sites with a 5ʹ "TTTN" PAM in the DNA strand opposite the target sequence (Supplementary Table S10). A highly active crRNA against ASFV p72 gene was selected from our previous study 37 .

In vitro RNA transcription using T7 RNA polymerase SARS-CoV-2 N and E genes were transcribed from the pBluescript-N and pUC57-E plasmids by adding a T7 promoter via PCR using Premix Taq (Cat. No. R004A, 15 TAKARA, Shuzo, Shiga, Japan). The crRNA templates were amplified from a pUC57-T7-crRNA (Supplementary Table S10 ) using a forward primer containing a T7 promoter and a reverse primer containing the nucleotide target sequences in different lengths (Supplementary Table S10 ). The PCR products were purified using the PCR purification Table S10 ). The Luna Universal Probe One-Step RT-qPCR Kit (Cat. No. E3006S, NEB) was used as the relevant master mix, RT-qPCR reactions were set up according to the manufacturer's instructions, and the thermocycling settings were according to the CDC and WHO protocol 49, 50 . N and E gene standards were used to generate a standard curve for copy number quantification. The qPCR assay and primers Table S10) 

Single-tube RAVI-CRISPR assay was conducted using components A and B. Component 

A single deep neural network, SSD (https://github.com/bubbliiiing/ssd-keras) 46 , was used to identify the images of PCR tubes, and the detection models were trained based on the Keras (v2.1.5) and TensorFlow (v1.13.1) frameworks 52, 53 . The images captured under no excitation light were used to train the PCR tube recognition models. Single-PCR tube and 8-tube PCR strip were used to improve the generalization power of the models. In The parameters for pooling layers were set as "pool_size = (2,2)". For fully connected layers, after 512-way rectified linear unit (ReLU) layers, a one-way sigmoid layer was designed to return a probability score. Additionally, we exposed the model to multiple aspects of the images through data augmentation, such as image rotation, image shift, and image shear, to enhance the generalization power. The parameters were: "rescale = 1./255, rotation_range = 40, width_shift_range = 0.2, height_shift_range = 0.2, shear_range = 0.2, zoom_range = 0.2, horizontal_flip = True". In addition, MagicEye mobile phone app (Android version 1.0) was developed and can be freely downloaded from "https://sourceforge.net/projects/ravi-crispr/files/MagicEye_en_v1.0.apk/download".

Statistical analysis was performed using R programming language. Mean ± standard errors of the mean (SEM) was determined for each treatment group in the separated experiments. The two-tailed Student's t test was used to determine significant differences 19 between treatment and control groups ( *** P < 0.001, ns: no significant at the 95% confidence level).

All study data are included in the article and Supplemental Information. 

Nucleic acid-based diagnostics for infectious diseases in public health affairs

Sensitive DNA detection and SNP discrimination using ultrabright SERS nanorattles and magnetic beads for malaria diagnostics

Advances in rapid detection methods for foodborne pathogens

A novel coronavirus outbreak of global health concern

Diagnostic Testing for Severe Acute Respiratory Syndrome-Related Coronavirus 2: A Narrative Review

COVID-19 diagnostic testing: Technology perspective

Considerations for diagnostic COVID-19 tests

Transmission routes of African swine fever virus to domestic pigs: current knowledge and future research directions

African Swine Fever Virus: A Review

The potential role of scavengers in spreading African swine fever among wild boar

African swine fever: A re-emerging viral disease threatening the global pig industry

PCR-based diagnostics for infectious diseases: uses, limitations, and future applications in acute-care settings

Point-of-care nucleic acid testing for infectious diseases

Detection of 2019 novel coronavirus (2019-nCoV) by real-time 21

Development of a TaqMan PCR assay with internal amplification control for the detection of African swine fever virus

Loop-mediated isothermal amplification of DNA

DNA detection using recombination proteins

A novel recombinase polymerase amplification (RPA) assay for the rapid isothermal detection of Neospora caninum in aborted bovine fetuses

Loop-mediated isothermal amplification (LAMP) of gene sequences and simple visual detection of products

Colorimetric detection of loop-mediated isothermal amplification reaction by using hydroxy naphthol blue

Rapid and visual detection of 2019 novel coronavirus (SARS-CoV-2) by a reverse transcription loop-mediated isothermal amplification assay

Single-copy sensitive, field-deployable, and simultaneous dualgene detection of SARS-CoV-2 RNA via modified RT-RPA

Detection of African swine fever virus by loop-mediated isothermal amplification

Rapid and Sensitive Recombinase Polymerase Amplification Combined With Lateral Flow Strip for Detecting African Swine Fever Virus

CRISPR-Cas12a based internal negative control for nonspecific products of exponential rolling circle amplification

Real-time kinetics and highresolution melt curves in single-molecule digital LAMP to differentiate and study specific and non-specific amplification

Nucleic acid detection with CRISPR-Cas13a/C2c2

CRISPR-Cas12a target binding unleashes indiscriminate singlestranded DNase activity

CRISPR-Cas12a-assisted nucleic acid detection

CDetection: CRISPR-Cas12b-based DNA detection with sub-attomolar sensitivity and single-base specificity

CRISPR-Cas12a has both cis-and trans-cleavage activities on singlestranded DNA

C2c2 is a single-component programmable RNA-guided RNA-targeting CRISPR effector

HOLMESv2: A CRISPR-Cas12b-Assisted Platform for Nucleic Acid Detection and DNA Methylation Quantitation

CRISPR-Cas12-based detection of SARS-CoV-2

CRISPR/Cas12a technology combined with immunochromatographic strips for portable detection of African swine fever virus

Cas12aVDet: A CRISPR/Cas12a-Based Platform for Rapid and Visual Nucleic Acid Detection

Application of CRISPR-Cas12a Enhanced Fluorescence Assay Coupled with Nucleic Acid Amplification for the Sensitive Detection of African Swine Fever Virus

Naked-Eye Detection of Grapevine Red-Blotch Viral Infection Using a Plasmonic CRISPR Cas12a Assay

Reverse Transcription Recombinase Polymerase Amplification 23

Coupled with CRISPR-Cas12a for Facile and Highly Sensitive Colorimetric SARS-CoV-2 Detection

CRISPR/Cas12a-Modulated fluorescence resonance energy transfer with nanomaterials for nucleic acid sensing

Deployable CRISPR-Cas13a diagnostic tools to detect and report Ebola and Lassa virus cases in real-time

Detection of SARS-CoV-2 with SHERLOCK One-Pot Testing

Amplification-free detection of SARS-CoV-2 with CRISPR-Cas13a and mobile phone microscopy

Ultrasensitive and visual detection of SARS-CoV-2 using all-in-one dual CRISPR-Cas12a assay

Contamination-free visual detection of SARS-CoV-2 with CRISPR/Cas12a: A promising method in the point-of-care detection

Single Shot MultiBox Detector

A sequence-specific plasmonic loopmediated isothermal amplification assay with orthogonal color readouts enabled by CRISPR Cas12a

CRISPR-offinder: a CRISPR guide RNA design and off-target searching tool for user-defined protospacer adjacent motif

CDC: Centers for Disease Control and Prevention. Real-time RT-PCR Panel for Detection 2019-nCoV

WHO: World Health Organization. Diagnostic Detection of Wuhan Coronavirus 2019 by Real-time RT-PCR (World Health Organization

African swine fever (infection with African swine fever) (World Organization for Animal Health

Deep learning with Keras: implement neural networks with Keras on Theano and TensorFlow

TensorFlow: A system for large-scale machine learning

A gene prioritization method based on a swine multi-omics knowledgebase and a deep learning model

Deep learning with Python

The authors declare no competing interests.