Digital PCR is an absolute quantitative analysis technology developed after real-time quantitative PCR. By transferring a single DNA molecule into an independent reaction chamber, the fluorescence signal was detected and analyzed after PCR amplification, and the absolute quantification of single DNA molecule was realized. Digital PCR technology has been widely used in gene mutation detection, copy number variation detection, microbial detection, genetically modified food detection and next generation sequencing. This paper introduces the quantitative method of digital PCR technology, and reviews the research progress of this technology in the main application fields.
The main principle of digital PCR is to place a single DNA molecule in an independent reaction chamber and amplify it by PCR. TaqMan chemical reagent and dye labeled probe are used to detect specific target sequences. The absolute quantification of samples is achieved by statistical analysis of the proportion and number of reaction units presenting two signal types. Therefore, digital PCR is also called single molecule PCR. Its detection process mainly includes two parts: PCR amplification and fluorescence signal analysis. Before the PCR reaction, the samples were dispersed into tens of thousands of units (reaction chambers), so that only a single DNA molecule existed in each unit. The amplification procedure and amplification system of PCR amplification phase are not different from ordinary PCR. In the phase of fluorescence signal analysis, terminal detection is used to collect the fluorescence signals of each reaction unit, and then directly count or borrow Poisson statistics to get the original concentration or content of the sample (Fig. 1). Unlike real-time fluorescence quantitative PCR, the whole process of digital PCR does not need to amplify standard curve and housekeeping gene, and has good accuracy and reproducibility, so it can achieve absolute quantification in real sense. Based on the different methods of liquid separation, digital PCR can be divided into three main types: microfluidic digital PCR (mdPCR), Droplet digital PCR (ddPCR) and chip digital PCR (cdPCR). Microfluidic channels, droplets or microfluidic chips are used to separate liquids. Each separated area can be separated by a single PCR reaction. The mdPCR based on microfluidic technology can separate DNA templates. Microfluidic technology can achieve nano-upgrading of samples or the formation of smaller droplets, but the droplets need special adsorption methods. Combined with the PCR reaction system, mdPCR has been gradually replaced by other methods; ddPCR technology is a relatively mature digital PCR platform, using oil-in-water micro-droplet generation technology. Current instruments mainly include Bio-rad's QX100/QX200 micro-droplet dPCR system and RainDance's RainDropTM dPCR system, including Bio-rad's dPCR system. The reaction system containing nucleic acid molecule is generated into 20,000 nano-upgraded droplets by water-in-oil generation technology. After PCR amplification, each droplet is detected by microdrop analyzer one by one. The preparation, reaction, separation and detection of samples are integrated into a chip by cdPCR using microfluidic chip technology. Current instruments mainly include Flu. Idigm's Bio-Mark gene analysis system and Life Technologies's QuantStudio system. Many microtubules and microcavities were engraved on silicon wafers or quartz glass by integrated fluid pathway technology, and the flow of solution was controlled by different control valves to realize liquid separation, mixing and PCR amplification of biological samples, so as to achieve absolute quantification.
The traditional PCR technology is to amplify the target gene by PCR amplification. A DNA template is duplicated into thousands of offspring and then detected by gel electrophoresis. However, the gel electrophoresis detection can only judge the molecular size of the amplified products, but can not deduce the content of DNA in the initial samples. Therefore, quantitative analysis can not be carried out. Real-time quantitative PCR can be used for absolute quantification and relative quantification, and absolute quantification is made of standard curves with a series of known concentrations. Under the same conditions, the fluorescence signal measured by the target gene was compared with the standard curve to obtain the quantity of the target gene. The standard sample could select purified plasmid DNA or ssDNA synthesized in vitro, and the relative content of DNA in the initial sample could be converted by internal parameters. Digital PCR is the third generation of PCR technology based on traditional PCR and real-time fluorescent quantitative PCR. It does not need standard products, nor does it need to make standard curves, that is, it can achieve more sensitive and accurate absolute quantification.
Another important factor in the high accuracy of digital PCR is the application of Poisson statistics to the detection of digital PCR. Because digital PCR is a terminal detection analysis method, if the target molecule is not well dispersed and there is more than one target DNA in a reaction chamber, then the concentration obtained is not credible. Poisson statistics is a descriptive method of random distribution. When the reaction chamber/droplet volume and the negative ratio are known, the initial concentration can be obtained by Poisson model. Therefore, if the reaction chamber is not saturated, the researchers can also calculate the number of starting molecules of the sample. Beer et al. have shown that the advantage of droplet-based method lies in its scalability. With the increase of the number of reaction containers, the quality of data increases accordingly. That is to say, when the number of reaction vessels is enlarged or increased, the Poisson accuracy will also increase.
At the same time, digital PCR can effectively avoid the influence of reaction inhibitors. As the reaction chamber increases, the reaction is less affected by the inhibitor.
Search the Web of Knowledge Platform Science Citation Index Expanded database for the keywords "digital PCR, dPCR" from 1999 to June 2016 (up to June 25, 2016). The search results show that there are 1 787 related articles, of which 1 405 articles of art type are analyzed by literature statistics. Figure 2 shows that the number of research papers related to digital PCR has increased year by year, and the number of papers has increased rapidly since 2010. Among many countries, the United States and China are the main countries in the research field of digital PCR (Figure 3). According to a study published by Kalorama Information, the global market for digital and qPCR is expected to reach US$3.97 billion in 2019, with China as the main emerging market.
In routine tissues and blood samples, it is difficult to detect single mutation because of its low somatic cell content. However, dPCR can dilute or partition the complex background with limited dilution, which can reduce the background signal of wild genotype, so that the low abundance target sequence can be detected sensitively, especially for the detection of rare mutations. Studies show that the mutation frequency as low as 1/100 000 can be detected by [5-6]. And with the increase of reaction chamber, the analysis of rare mutation is more sensitive.
DPCR has a wide range of applications and research in rare mutation detection, especially in cancer-related detection and quantitative research, such as EGFR , BRAF , KRAS , PIK3CA [10-11], JAK2  and microRNAs . PIK3CA can be activated by growth factors such as IGF-1, HGF and EGF. It acts with growth factors on receptor tyrosine kinase, promoting proliferation, angiogenesis and cell metabolism. Kim et al. used ddPCR to detect the PIK3CA mutation of free DNA in serum . The low abundance PIK3CA mutation in serum was successfully detected. Thirty-eight metastatic cholangiocarcinoma samples were tested. One matched serum sample PIK3CA p.E542K was positive for 28 mutant copies, which was equivalent to the prevalence of 48 copies/mL serum and 0.3% allele. Another serum sample PIK3CA p.H1047R was positive for 10 mutant copies, which was equivalent to 18 copies/mL serum and 0.2% allele. The rate of disease. Miotto et al showed that EvaGreen dye method and TaqMan probe method could be used to detect circulating microRNAs in human plasma and serum by ddPCR.
Usually, gene expression analysis relies on agarose gel electrophoresis, realtime PCR and other methods, and the determination of copy number is always limited. The duplex reaction of target gene and reference gene was counted directly by dPCR. By calculating the ratio, the copy number of target gene [15-17] was obtained directly.
The discovery of Copy Number Variations (CNV) of CCL3 coding genes affecting HIV infection and disease progression has aroused widespread controversy and has been partly attributed to the different results obtained by the methods of evaluating the number of copies. CCL3 gene encodes CC chemokine CCL4, which is also the natural ligand of CCR5 receptor, and plays a protective role on HIV-1. Bharuthram and other standard methods qPCR and ddPCR were used to evaluate the CCL4L gene copy number of . The results showed that compared with qPCR (r = 0.87, P < 0.0001), the sum of CCL4L copies and CCL4L 1 and CCL4L2 copies measured by ddPCR had a good correlation (r = 0.99, P < 0.0001). However, the accuracy of qPCR at high copy number decreased significantly. Whale et al. simulated the HE2 gene amplification of different CNVs. Under the same experimental conditions, dPCR could detect lower CNVs  than qPCR. Because the accuracy of dPCR is directly related to small amplification and template concentration, they have also established a dPCR method for measuring CNV. The variance expression of dPCR is obtained by Poisson and two item distribution.
Nucleic acid amplification (NAA) is an important method for pathogenic microorganism detection. However, there may be a problem that sometimes the presence of low-level target molecules and small molecular inhibitors may lead to the deviation of quantitative results. DPCR is a detection method that does not require standard curve and is insensitive to some inhibitors. Therefore, researchers have applied it to the detection of some viruses or pathogenic bacteria, such as HBV [20-21], enterovirus , adeno-associated virus , cytomegalovirus , methicillin-resistant Staphylococcus aureus , Mycobacterium tuberculosis , Shigella toxin-producing Escherichia coli . Strain et al. used dd-PCR and q-PCR to detect residual DNA in HIV patients treated with combined antiretroviral therapy. The study found that dd-PCR had higher accuracy and lower copy variation coefficient . Dong et al. established ddPCR for the rfbE gene of E.coli O157:H7 and optimized the probe concentration. In a 20-mu-L system, the detection range of genomic DNA of E.coli O157:H7 was 4_1.25*105copies, and the linear correlation coefficient was 0.999. At the same time, 16 samples were tested by ddPCR and qPCR, and the same results were obtained.
In environmental detection, dPCR has also been applied to [30-31]. Josefa et al. found that dPCR was less sensitive to reaction inhibitors than qPCR in qualitative and quantitative analysis of low-copy DNA of Phytophthora tobacco in soil and plant rhizome samples. Because of the deviation of realtime PCR based on standard curve in detection, Cao et al. used double DD PCR to simultaneously detect Enterococcus and human feces related marker HF183  in fecal source and water quality. Dual Enterococci and HF183 compete with each other in qPCR, which makes them undetectable or beyond their single minimum detection concentration range, while the single and double detection results of ddPCR are consistent. Its tolerance, repeatability, sensitivity and accuracy to inhibitors were higher than those of qPCR, but its quantitative upper limit was lower than that of realtime PCR. Anja et al. combined dd-PCR with q-PCR, used dd-PCR to prepare standard curve and real-time PCR to quantify Listeria monocytogenes in different biofilm formation stages.
European Union countries have strict restrictions on genetically modified foods, but at present, real-time quantitative PCR is generally used for molecular quantitative analysis of genetically modified foods in commodities. But in some complex food and feed raw materials, the target DNA is very few, which limits its detection and quantification. DPCR provides a new method for researchers both at home and abroad. Morisset et al. Used ddPCR to make absolute quantitative  for MON810 transgenic and HMG maize reference genes. Studies have shown that ddPCR has higher repeatability for detection of low concentration and higher tolerance to inhibitors. Dobnik et al. used multiplex dd-PCR to detect 12 EU-authorized GM maize. For samples containing GM tissues, this method has higher throughput and higher efficiency than real-time quantitative PCR . Damira et al. established a dual-dd-PCR to detect the proportion of T-nos/hmg gene copies in quantitative transgenic maize. Through central composite design optimization, DNA digestive enzymes were added to the system to reduce detection bias. The detection limit of T-nos was 11 copies, and the dynamic range of T-nos/hmg ratio was 0.08%-100%.
Next-generation sequencing (NGS), also known as high-throughput sequencing, is a revolutionary advance in sequencing technology. It can sequence millions of DNA molecules at a time, making it possible to analyze the transcripts and genomes of a species in detail. The dPCR platform can be docked with NGS to control the quality of sequencing libraries and provide quantitative analysis and quality evaluation information for sequencing libraries. On the one hand, dPCR validates the sequencing results of NGS, verifies genomic variations such as single nucleotide polymorphisms, mutations and copy number variations to ensure the reliability of sequencing results; on the other hand, the results also contain information reflecting the quality of sequencing libraries, such as junction and junction dimers, misconnection. Segment, too long connection fragments, etc., this is the other methods do not have the advantage. Alikian et al. showed that dPCR was more accurate than qPCR in the detection of chronic myeloid leukemia patients, and the quantitative analysis results were reliable and reproducible.
Compared with common PCR and qPCR, dPCR has unique technical advantages and achieves single molecule DNA absolute determination.