Method and system for high-throughput screening of antibacterial phages

Abstract

A method and system for high-throughput screening of antibacterial phages, the method comprising: 1) spotting a plurality of phages to be detected on a plate containing host bacteria and culturing; 2) photographing the cultured plate to obtain a plaque image; 3) binarizing the plaque image to convert the phage image into a binary image; 4) filtering the binary image based on the shape and transmittance of the plaques in the binary image to obtain phage antibacterial phenotype data; 5) screening the phages to be detected based on the phage antibacterial phenotype data to obtain the antibacterial phages. The method and system are used for identifying phage infectivity, selecting phage monoclonal, preparing a phage cocktail preparation for phage therapy, phage function verification, reverse screening of phage-resistant host bacteria, and phage lysozyme function verification.

Description

Technical Field

[0001] The present invention belongs to the field of biotechnology. More specifically, the present invention provides a method and a system for high-throughput screening of phages. Background Art

[0002] A phage is a virus that specifically infects bacteria and is widely present in places where bacterial communities are distributed, such as soil, the ocean, the intestines of humans and animals, etc. A phage does not have a cell structure and is mainly composed of a protein coat and a single nucleic acid DNA or RNA genetic material encapsulated therein. The length of a phage ranges from 20 nanometers to 200 nanometers, and its genome can encode as few as several or as many as hundreds of genes. A phage cannot grow or replicate independently and must utilize the energy and metabolic systems in the host bacteria to achieve its own growth and proliferation. A phage recognizes the host by specifically binding to receptors on the surface of the bacteria, and thus has strict host specificity.

[0003] As early as the 1920s, at the very beginning of the discovery of phages, the discoverer of phages, Felix d’Herelle, used phages for clinical applications of bacterial infections, such as for preventing and controlling the outbreaks of cholera in India and plague in Egypt. However, due to the insufficient basic research on phages at that time, there were problems with phage therapy, such as inconsistent efficacy evaluation criteria, difficult standardization of production methods, and low product purity. With the discovery and use of antibiotics, phage therapy was soon replaced by the cheaper and more efficient antibiotic therapy. Phage therapy thus gradually faded out of the medical and research systems in Western developed countries. In recent years, with the spread of antibiotic-resistant bacteria globally, the therapeutic effect of antibiotics on bacterial infections has been severely challenged, which has also prompted some scientists to re-invest in the research of phage therapy. In July 2005, the journal Antimicrobial Agents and Chemotherapy reported the first standardized randomized double-blind human trial of phage therapy, demonstrating the safety of orally administered phage preparations for humans. In June 2009, the journal Journal of Wound Care reported the first Phase I clinical trial approved by the US FDA, demonstrating the safety of phage preparations in wound treatment. The demand for clinical applications and the development of related technologies have re-guided the continuous development of the phage field. With the rapid development of next-generation sequencing technology, the number of phages isolated from natural environments has been increasing day by day, and the genomic data has also been increasing. However, the relevant protein annotations are often limited by the lack of database functional verification information and are not sufficient to support the future trend of phage research. The bottleneck restricting the functional verification of phages lies in the low efficiency, high manual dependence, and high cost of traditional technical methods, especially the time cost.

[0004] The clinical application promotion of phage therapy, which uses phages to infect and eliminate bacteria, first requires determining the host infection range (host spectrum). Traditional phage host spectrum identification mostly uses plaque assays (double-layer agar plate method) with semi-solid media. Compared with spot or cross-streak assays, the double-layer agar plate method has higher quantifiability and is a conventional and most widely accepted method. The plaque assay was initially established by Felix d’Herelle, one of the discoverers of phages, and later improved by many other phage biologists. The plaque assay observes the clear areas in bacterial plates formed by the amplification of single phage particles. The size and the presence or absence of halos are related to their own infectivity. Although the plaque assay is simple, the morphology and size of plaques may vary with the experimenter, culture medium, and other conditions. Moreover, the double-layer agar plate method is laborious, time-consuming, expensive, requires a large number of plates and manual work, cannot perform large-scale high-throughput assays, and cannot quantitatively study phage infectivity. The quantitative description of phage infectivity by the double-layer agar plate method is subject to the subjective influence of the experimenter and cannot achieve absolute quantification only through subjective descriptions such as "large, strong, small, average".

[0005] Other methods for determining the phage host spectrum include liquid lysis assays. When phages infect hosts, grow themselves, and lyse the hosts, the infectivity of phages can be reflected by monitoring the decrease in the optical density of bacterial cultures. The Bioscreen C analyzer automatically lyses and determines the infectivity of phages by this method. The disadvantage of this method is that the bacterial cell debris generated by lysis may affect the recorded optical density value, thus possibly masking or underestimating the true phage lysis activity. Additionally, the lysis activity can be indirectly determined by tracking the decrease in the optical density of bacterial cultures due to liquid lysis. For example, the OmniLog TM system uses an automated, high-throughput, indirect liquid lysis assay method to evaluate the lysis activity of phages. OmniLog TMThe system uses redox chemical reactions and utilizes cell respiration as a universal reporter gene. During active bacterial growth, cell respiration reduces tetrazolium dyes and produces color changes that are measured automatically. Phage infection and subsequent phage growth in its host bacteria lead to reduced bacterial growth and respiration and directly cause a corresponding decrease in color. Different from the above two methods, the growth of phages in a given bacterial strain can be measured more precisely through a one-step growth experiment, providing data on each step in the phage life cycle, such as adsorption efficiency, latent period, eclipse period, burst size, etc. However, the one-step growth experiment not only consumes resources and reagents (e.g., requires a large number of plates), but is also labor-intensive and thus impractical for high-throughput analysis. Moreover, whether it is observing the change in OD value of bacterial growth by liquid lysis, or measuring phage one-step growth experiment, or monitoring redox reactions and using cell respiration as a universal reporter gene to indirectly measure live host bacteria, these methods cannot accurately quantify the ability of phages to infect bacteria. These methods have a narrow scope for functional verification and cannot effectively measure functions and characteristics related to phage infection. For example, halos are formed due to reasons such as phage self-lytic enzymes, and this apparent feature cannot be effectively observed in a liquid verification system, thus unable to cover the in-depth requirements for phage-related functional verification and quantitative analysis.

[0006] Therefore, there is currently no efficient and systematic method for high-throughput determination of the ability of phages to infect host bacteria, and it is difficult to quantitatively analyze phage antibacterial phenotypes. This problem needs to be overcome in both the basic scientific research field of phages and the popularization and application of phage therapy. Summary of the Invention

[0007] In view of this, the purpose of the present invention is to develop an efficient high-throughput method and system for screening antibacterial phages in combination with an efficient enhanced image recognition algorithm, aiming at the existing technical defects such as low efficiency and poor accuracy, and it can be particularly applied to aspects such as phage host spectrum identification, function recognition, and infection efficiency determination.

[0008] Therefore, on the one hand, the present invention provides a method for high-throughput screening of antibacterial phages, the method comprising:

[0009] 1) Spotting multiple phages to be detected on a plate containing host bacteria and culturing them;

[0010] 2) Taking an image of the cultured plate to obtain a plaque image;

[0011] 3) Binarizing the plaque image to convert the phage image into a binary image;

[0012] 4) Filter the binary image based on the shape and transmittance of the plaque in the binary image to obtain phage antibacterial phenotype data;

[0013] 5) Screen the phages to be detected based on the phage antibacterial phenotype data to obtain the antibacterial phages.

[0014] In one embodiment, in 1), the plate is a double-layer agar plate.

[0015] In one embodiment, in 1), the multiple phages to be detected are spotted in an array on a double-layer agar plate containing host bacteria.

[0016] In one embodiment, in 1), using a mechanical application force or an ultrasonic application force, aspirate or dip the phage liquid of multiple host spectra to be detected and spot it on a plate containing host bacteria.

[0017] In one embodiment, in 2), the phage image is a grayscale image; or the phage image is a color image, and the color image is converted to a grayscale image before being converted to a binary image.

[0018] In one embodiment, in 2), enhance the plaque image to eliminate the influence of uneven illumination on the image.

[0019] In one embodiment, use the Retinex algorithm to enhance the plaque image.

[0020] In one embodiment, enhance the plaque image by the following steps:

[0021] (a) The obtained original image S(x, y) is expressed as the product of the illumination image L(x, y) and the reflectance image R(x, y), as shown in Equation (1):

[0022] S(x, y) = R(x, y) * L(x, y) (1);

[0023] Taking the logarithm of both sides of Equation (1) gives:

[0024] log[R(x, y)] = log[S(x, y)] – log[L(x, y)] (2);

[0025] (b) Approximately obtain the illumination image L(x, y) by the convolution (Gaussian filtering) of the original image S(x, y) and a Gaussian kernel G(x, y), and convert Equation (2) to obtain Equation (3);

[0026] log[R(x, y)] = log[S(x, y)] – log[S(x, y) * G(x, y)] (3);

[0027] (c) Quantize the obtained log[R(x,y)] result into pixel values in the range of [0, 255], and then output the result image R(x,y):

[0028] R(x,y) = (log[R(x,y)] – Min) / (Max – Min) * 255 (4);

[0029] Wherein, Min is the minimum value in log[R(x,y)], and Max is the maximum value in log[R(x,y)].

[0030] In one embodiment, in 3), fuse two sizes to binarize the plaque image.

[0031] In one embodiment, in 4), the shape of the plaque in the binary image refers to the circularity of the plaque, which is calculated by the following formula (5):

[0032] Circularity = 4×π×area÷(perimeter×perimeter) (5),

[0033] Wherein, the area and perimeter are respectively the area and perimeter of the plaque in the binary image.

[0034] In one embodiment, the filtering is to remove plaques with a circularity lower than 0.1.

[0035] In one embodiment, in 4), the transmittance of the plaque in the binary image is calculated by the following formula (6),

[0036] Transmittance of plaque = plaque brightness value / background brightness value (6).

[0037] In one embodiment, the filtering is to remove plaques with a transmittance greater than 0.995.

[0038] In one embodiment, in 5), use the shape and transmittance of the plaque to calculate a comprehensive score, make a plaque judgment according to the comprehensive score, and screen the phage to be detected based on the plaque judgment result to obtain the antibacterial phage, wherein the comprehensive score is calculated by the following formula (7):

[0039] Comprehensive score = circularity + 20×(1 - transmittance of plaque) (7).

[0040] In one embodiment, plaques with a comprehensive score greater than 1 are judged as positive plaques, and the phage to be detected corresponding to the positive plaques is an antibacterial phage.

[0041] In a second aspect, the present invention provides a high-throughput screening system for antibacterial phages, and the system includes:

[0042] A phage culture unit for spotting multiple phages to be detected on a plate containing host bacteria for culturing;

[0043] An image acquisition unit for taking an image of the cultured plate to obtain a plaque image;

[0044] An image binary conversion unit for converting the phage image into a binary image;

[0045] A data extraction unit for filtering the binary image based on the shape and transmittance of the plaques in the binary image to obtain data on the antibacterial phenotype of the phage;

[0046] An analysis unit for screening the phages to be detected based on the phage antibacterial phenotype data to obtain the antibacterial phages.

[0047] In one embodiment, the system further includes:

[0048] An image enhancement unit for enhancing the plaque image to eliminate the influence of light intensity on the image.

[0049] In one embodiment, in the phage culture unit, the plate is a double-layer agar plate.

[0050] In one embodiment, in the phage culture unit, the multiple phages to be detected are spotted in an array on a double-layer agar plate containing host bacteria.

[0051] In one embodiment, in the phage culture unit, a mechanical force or an ultrasonic force is used to aspirate or dip a phage liquid of multiple host spectra to be detected and spot it on a plate containing host bacteria.

[0052] In one embodiment, the phage image obtained by the image acquisition unit is a grayscale image; or the image acquisition unit obtains a color phage image and converts the color image into a grayscale image before converting it into a binary image.

[0053] In one embodiment, in the image acquisition unit, the plaque image is enhanced to eliminate the influence of uneven illumination on the image.

[0054] In one embodiment, the Retinex algorithm is used to enhance the plaque image.

[0055] In one embodiment, the following steps are used to enhance the plaque image:

[0056] (a) The obtained original image S(x, y) is expressed as the product of the illumination image L(x, y) and the reflectance image R(x, y), as shown in Equation (1):

[0057] S(x, y) = R(x, y) * L(x, y) (1);

[0058] Taking the logarithm of both sides of Equation (1) gives:

[0059] log[R(x, y)] = log[S(x, y)] – log[L(x, y)] (2);

[0060] (b) The illumination image L(x, y) is approximately obtained by convolving the original image S(x, y) with a Gaussian kernel G(x, y) (Gaussian filtering), and Equation (2) can be transformed into Equation (3);

[0061] log[R(x, y)] = log[S(x, y)] – log[S(x, y) * G(x, y)] (3);

[0062] (c) The obtained result of log[R(x, y)] is quantized into pixel values in the range of [0, 255], and then the result image R(x, y) is output, as shown in Equation 4:

[0063] R(x, y) = (log[R(x, y)] – Min) / (Max – Min) * 255 (4);

[0064] where Min is the minimum value in log[R(x, y)], and Max is the maximum value in log[R(x, y)].

[0065] In one embodiment, in the image binary conversion unit, two sizes are fused to binarize the plaque image.

[0066] In one embodiment, in the data extraction unit, the shape of the plaque in the binary image refers to the circularity of the plaque, which is calculated by the following formula (5):

[0067] Circularity = 4 × π × area ÷ (perimeter × perimeter) (5),

[0068] where the area and perimeter are the area and perimeter of the plaque in the binary image, respectively.

[0069] In one embodiment, the filtering is to remove plaques with a circularity lower than 0.1.

[0070] In one embodiment, in the data extraction unit, the transmittance of the plaque in the binary image is calculated by the following formula (6),

[0071] Transmittance of plaque = Brightness value of plaque / Brightness value of background (6).

[0072] In one embodiment, the filtration is to remove plaques with a transmittance greater than 0.995.

[0073] In one embodiment, in the data extraction unit, a comprehensive score is calculated using the shape and transmittance of the plaque, plaque judgment is performed based on the comprehensive score, and the phage to be detected is screened based on the plaque judgment result to obtain the antibacterial phage, wherein the comprehensive score is calculated by the following formula (7):

[0074] Comprehensive score = Circularity + 20×(1 - Transmittance of plaque) (7).

[0075] In one embodiment, plaques with a comprehensive score greater than 1 are judged as positive plaques, and the phage to be detected corresponding to the positive plaques is the antibacterial phage.

[0076] In one embodiment, after it is determined as a positive plaque, when the difference in the comprehensive score is less than 1, it is determined that the antibacterial ability tends to be the same or similar. For example, when the comprehensive score is greater than 1 and divided by each integer interval, positive plaques with a comprehensive score between 1 - 2 (excluding 2) are determined to have similar antibacterial abilities, and positive plaques with a comprehensive score between 2 - 3 (excluding 3) are determined to have similar antibacterial abilities... and so on.

[0077] In a third aspect, the present invention provides a method for identifying phage infectivity, selecting phage monoclonal, formulating a phage cocktail preparation for phage therapy, verifying phage function, reverse screening for phage-resistant host bacteria or verifying phage lyase function, the method comprising screening antibacterial phages using the high-throughput screening method for antibacterial phages according to the first aspect of the present invention.

[0078] In one embodiment, antibacterial phages are screened using the high-throughput screening system for antibacterial phages according to the second aspect of the present invention.

[0079] In one embodiment, the method for the phage cocktail preparation based on phage therapy comprises:

[0080] 1) Screening effective phages that target and infect the target bacteria and quantitatively analyzing antibacterial infection characteristics, and screening candidate phages according to the phage cocktail preparation standard;

[0081] 2) Amplifying and culturing the candidate phages and formulating a phage cocktail preparation;

[0082] 3) Evaluating the effect of the phage cocktail preparation on the target bacteria.

[0083] In one embodiment, step 1) is carried out using the system for high-throughput screening of antibacterial phages according to the second aspect of the present invention.

[0084] In one embodiment, the formulation criteria for the cocktail preparation include: 1) being only effectively bactericidal against the target bacteria and not effectively bactericidal against other bacteria; 2) being able to effectively exert its effect and effectively reduce the off-target effect; 3) the host spectrum ranges of different phages being complementary, that is, the bacterial resistance mutants generated by one phage being sensitive to another phage.

[0085] In one embodiment, the matrix in the preparation is 0.9% NaCl solution, and the phage titer is 10^9 (pfu / ml).

[0086] In one embodiment, the phage function verification includes the verification of the functions of the phage's own genes and the functions of the bactericidal / bacteriostatic substances of the phage.

[0087] The present invention provides a high-throughput protocol for quantitative analysis of the antibacterial phenotype of phages, distributing a reasonable number of phages on the limited semi-solid culture plate space to achieve high-throughput detection of phage hosts. In addition, the present invention also performs quantitative analysis of plaques by using an enhanced image recognition algorithm, and analyzes the phage infection ability based on the phenotypic characteristics such as the radius and shape parameters formulated according to the biological characteristics of the plaques, so as to screen and obtain antibacterial phages. The method and system of the present invention are suitable for being made into an integrated system device. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for description in the embodiments or the prior art.

[0089] Figure 1 Shows a schematic diagram of enhancing the plaque image using the Retinex algorithm.

[0090] Figure 2 Shows a schematic diagram of binarizing the plaque image by fusing two sizes.

[0091] Figure 3 Shows the filtering of the binary image using the shape and transmittance of the plaques in the binary image.

[0092] Figure 4 Shows a heat map of the scoring of the infection ability of multiple phages against multiple bacteria, and visually displays the phage infection ability using software.

[0093] Figure 5 Shows an exemplary overall structure of the system for high-throughput screening of antibacterial phages and a comparison of the efficiency with the process of identifying phages by traditional methods.

[0094] Figure 6 Shows the results of high-throughput phage infection of host bacteria measurement based on automated mechanical implementation force and image enhancement recognition.

[0095] Figure 7 Shows the results of high-throughput phage infection of host bacteria measurement based on ultra-micro sound wave implementation force and image enhancement recognition, taking Acinetobacter baumannii "1050" as an example.

[0096] Figure 8 Shows the results of high-throughput phage infection of host bacteria measurement based on ultra-micro sound wave implementation force and image enhancement recognition, taking Klebsiella pneumoniae "9384" as an example.

[0097] Figure 9 Shows the influence of image enhancement recognition algorithm on identifying plaque areas.

[0098] Figure 10 Shows the bactericidal effect of X3 preparation against the target host bacteria.

[0099] Figure 11 Shows the one-step growth curve of phage taking the natural phage PHAB8962 as the experimental sample.

[0100] Figure 12 Shows the bactericidal curves at different titers taking the natural phage PHAB8962 as the experimental sample. Detailed implementation mode

[0101] The present invention overcomes the related technical drawbacks faced in phage function verification and quantitative analysis, and provides a complete high-throughput method and system for screening antibacterial phages, especially in the identification of the host bacteria spectrum. Using the method and system of the present invention can reduce the identification cost, save labor costs, improve the efficiency of quantitative determination, and contribute to the output of relevant biological characterization data, while not losing the accuracy of the determination. The method and system of the present invention can achieve a reasonable distribution of phages in a limited semi-solid culture plate space at an ultra-trace level by using methods such as mechanical force and acoustic force, so as to achieve high-throughput monitoring of phage infection of host bacteria at a limited space level, greatly improving the efficiency and reducing the cost. The method and system of the present invention further utilize phage plaque image algorithms to enhance recognition and quantitative analysis, and based on phenotypic characteristics such as the radius and shape parameters formulated according to phage biological characteristics, guide in-depth analysis of phage infection ability, and measure relevant quantitative physical characteristics. In addition, based on a large amount of phage data, the present invention sets reasonable screening thresholds, and finally forms an overall complete phage function verification, host spectrum identification and recognition, and quantitative analysis integrated system solution. The method and system of the present invention solve the phage characterization analysis as a whole from the source to the analysis of phenotypic data output, focus on overcoming the problems of low efficiency and high error in host spectrum identification and recognition, and can be transformed into devices and equipment.

[0102] In the present invention, high-throughput phage infection of host bacteria can be achieved through automated integrated implementation of force. A large number of phages are quantitatively spotted onto a double-layer agar plate containing host bacteria to form an array of phages. High-throughput phage quantitative spotting can be achieved through automated mechanical implementation of force or ultra-trace acoustic implementation of force for host bacteria infection. Using mechanical implementation of force (such as an automated pipetting platform, such as Hamilton, multi-channel pipetting equipment), prepare and pour a double-layer agar plate containing host bacteria, and constantly pipette or dip the phage liquid to be detected for the host spectrum (which can be placed in a 96-well, 384-well plate, etc.) onto the double-layer agar plate. Air-dry the double-layer agar plate and culture it under suitable growth conditions such as appropriate temperature. Finally, under stable light source, focal length and other camera parameters, perform unified image shooting to obtain relevant image data. Using acoustic implementation of force (such as an ultra-trace pipetting device Echo550), and constantly pipette the phage liquid to be detected for the host spectrum (which can be placed in a 96-well, 384-well plate, etc.) onto the prepared and poured double-layer agar plate containing host bacteria. Air-dry the double-layer agar plate and culture it under suitable growth conditions such as appropriate temperature. Finally, under stable light source, focal length and other camera parameters, perform unified image shooting to obtain relevant image data. The shooting can use a digital camera to obtain digital images.

[0103] In the present invention, accurate identification of plaque is achieved by enhancing image recognition. The Retinex algorithm can be used to enhance the plaque image. Due to factors such as lighting and materials, the contrast between the plaque and the surrounding halos and the background is low, which is not conducive to subsequent recognition and segmentation. Therefore, the present invention uses the Retinex algorithm to enhance the plaque image, Figure 1 (including Figure 1 A-1C) shows the principle and exemplary results of enhancing the plaque image.

[0104] As Figure 1 shown in A, the observed image S is obtained by the reflection of the incident light L on the object surface, and the reflectivity R is determined by the object itself and is not affected by the change of the incident light L. The original image S(x, y) here is the product of the illumination image L(x, y) and the reflectivity image R(x, y), and can be expressed as the following formula:

[0105] S(x, y) = R(x, y) * L(x, y) (1),

[0106] where (x, y) represents coordinates, L(x, y) represents the incident image, which determines the dynamic range that the image pixels can reach, S(x, y) represents the original image, and R(x, y) represents the reflection property of the object, that is, the inherent attribute of the image, which is the term to be obtained;

[0107] Taking the logarithm of both sides of formula (1), the original information of the object can be obtained:

[0108] Log[R(x, y)] = log[S(x, y)] – log[L(x, y)] (2),

[0109] Approximating the illumination image L(x, y) by the convolution (Gaussian filtering) of the original image S(x, y) and a Gaussian kernel G(x, y), formula (2) can be converted to formula (3);

[0110] log[R(x, y)] = log[S(x, y)] – log[S(x, y) * G(x, y)] (3);

[0111] Quantifying the obtained result of log[R(x, y)] into pixel values in the range of [0, 255], and then outputting the result image R(x, y) (exemplary results are shown in 1B and 1C), as shown in formula 4:

[0112] R(x, y) = (log[R(x, y)] – Min) / (Max – Min) * 255 (4);

[0113] where Min is the minimum value in log[R(x, y)], and Max is the maximum value in log[R(x, y)].

[0114] Image enhancement can be achieved through computer software and hardware, for example, through a computer program, such as a program written in Python. Therefore, image enhancement can be achieved through an image enhancement module, which is used to enhance the plaque image and eliminate the influence of light intensity on the image.

[0115] In the present invention, the plaque image is binarized by fusing two sizes (50 pixels and 31 pixels are selected through comparative experiments). Specifically, after the image is enhanced, the enhanced image is converted from a color image to a grayscale image. Then, the grayscale image is converted into a binary image (the plaque area is extracted). Figure 2 (including Figure 2 A and 2B) show a schematic diagram of binarizing the plaque image by fusing two sizes. The overall brightness of the plaque image fluctuates greatly and the brightness distribution is uneven. Therefore, only local adaptive threshold binarization can be used. The size range of the plaque is relatively large. Therefore, using a single size for binarization is likely to cause omission or misrecognition. When the radius R of the search window is small, some large spots will form hollows, resulting in omission. When the search radius R is large, it is likely to cause misrecognition and identify impurities as spots. Therefore, the present invention uses two search window sizes to binarize the image respectively, Figure 2 A shows the recognition map of the smaller size, Figure 2 B shows the recognition map of the larger size. Then, based on the recognition map of the smaller size and combining the spots newly recognized in the larger size map, the two binarized maps are fused to reduce omission and misrecognition.

[0116] The binarization of the plaque image can be achieved through computer software and hardware, for example, through a computer program. Therefore, the binarization of the plaque image can be achieved through a binary image conversion module, which is used to binarize the plaque image enhanced by the image enhancement module, convert the phage image from a color image to a grayscale image, and convert the grayscale image into a binary image.

[0117] In the present invention, recognition features and threshold filtering are also performed. Although fusing two sizes for binarization can reduce the probability of misrecognition, there is still a certain probability of misrecognition. Therefore, the binarized map is further filtered by using the shape and transmittance of the plaque in the binarized map.

[0118] Shape parameter: Since most plaques are circular or elliptical, while the shapes of plaques or impurities with low contrast generally appear irregular. Therefore, the inventor takes the circularity of the plaque as one of the plaque credibility parameters, and the inventor uses formula (5) to calculate the circularity of the plaque:

[0119] Roundness = 4×π×Area÷(Perimeter×Perimeter) (5)

[0120] The area and perimeter here are respectively the area and perimeter of the plaque in the binary image. The perimeter of the target area, taking Figure 3 A as an example, the perimeter of the area in the figure is the length of the outer circle line (in pixels). The area of the target area is the area of the white area (in pixels). The higher the shape parameter, the rounder the target area. The value generally ranges from 0 to 1. When the shape parameter is lower than 0.1, the target area is regarded as an impurity.

[0121] Transmittance parameter: Since the plaque is generally darker (compared to the surrounding area), the inventor takes the transmittance as one of the credibility parameters of the plaque. First, detect the plaque and its surrounding halo, such as Figure 3 the inner circle and outer circle in B. Then, measure the up-down and left-right ranges of the plaque, that is, Figure 3 diameter_x and diameter_y in B. Using the plaque (inner circle area) as a marker, expand diameter_x and diameter_y to the left and right and up and down respectively as the background preselection area, and then remove all plaques and halos (including other possibly enclosed plaques and halos) within this area, that is, Figure 3 the area outside the outer circle in C is used as the reference area for background analysis. Then, take the median of the pixel brightness values (converted to grayscale values) within the inner circle area as the plaque brightness value, and take Figure 3 the median of the inner pixel brightness values (converted to grayscale values) in the area outside the outer circle in C as the background brightness value. Finally, calculate the transmittance of the plaque,

[0122] Transmittance of the plaque = Plaque brightness value / Background brightness value (6)

[0123] Generally, the transmittance is 0.5 - 1. The lower the transmittance, the darker the target area and the greater the possibility of it being a plaque. Preferably, plaques with a transmittance greater than 0.995 can be considered as impurities and need to be removed.

[0124] Finally, a comprehensive score can also be calculated. Since the range of the transmittance is 0.5 - 1, the lower the transmittance, the greater the probability that the target is a plaque, and the transmittance has a stronger guiding significance for the determination of the plaque. Therefore, the formula for the comprehensive score of the final target area is:

[0125] Comprehensive score = Roundness + 20×(1 - Transmittance of the plaque) (7)

[0126] In the tests conducted by the inventors, it was found that all the images with a comprehensive score higher than 1 were experimentally verified as positive plaques. Among the images with a comprehensive score in the range of (0, 1), 15% were experimentally verified to have a probability of being positive plaques. All the images with a comprehensive score less than or equal to 0 were experimentally verified as negative plaques. Therefore, it is possible to determine whether a plaque is genuine based on the comprehensive score. If it is genuine, it is a positive plaque, and it is determined that the phage corresponding to the positive plaque can infect the bacterium, and this phage is a phage resistant to the bacterium. The level of the comprehensive score can relatively represent the strength of the phage infection ability. At the same time, the measured plaque radius, area, circularity, halo length, etc. can be output as quantitative analysis data for phage characterization and for verifying specific functions such as phage infection and replication. Then, for different antibacterial application directions, further quantitative analysis of single-phenotype data is carried out based on each single phage antibacterial phenotype (such as transmittance, radius, etc.). Phages with large plaque areas and transparency (low transmittance) are screened, and usually have stronger antibacterial effects. In specific antibacterial applications, several phages with the highest comprehensive scores can be selected. When the comprehensive scores are close (for example, the difference is less than 1), it can be determined that the antibacterial abilities of the phages tend to be the same or similar. For example, for the comprehensive score of phages against target bacteria, when the comprehensive score is greater than 1, the abilities of phages against target bacteria can be grouped according to each integer interval.

[0127] Filtering binary images by shape and transmittance can be achieved through computer software and hardware, such as through a computer program. Therefore, filtering binary images by shape and transmittance can be implemented by an image filtering module, and the image filtering module is used to further filter binary images based on the shape and transmittance of plaques in the binary images.

[0128] In the present invention, phenotypic deep quantitative analysis and visualization of image recognition data can also be further performed based on the positive screening results of phages. For example, by integrating and analyzing the data of plaques accurately identified by an enhanced image recognition algorithm, heat map analysis can be performed to compare the phage infection abilities of the population (see Figure 4 ), correlation analysis can be used to screen specific-purpose phages (such as Examples 1 and 2); high-throughput screening and verification of combined phage cocktail preparations (such as Example 4); verification of specific functions of natural or synthetically modified phages (such as Example 5); reverse screening of phage-resistant host bacteria; verification of lyase-related functions (relying on the size of the outer halo), etc., ultimately realizing phage-related applications.

[0129] Figure 5 The overall structure of an exemplary high-throughput screening system for antibacterial phages is shown, as well as a comparison with the process of identifying phages by the traditional double-layer agar plate method. According to Figure 5As shown, the traditional manual double-layer agar plate method is limited by low manual precision and plate throughput. It can detect a throughput of 24 wells in a relatively ideal state and cannot quantify various phenotypic characteristics. The system of the present invention has been experimentally verified that both the embodiment based on mechanical force or ultrasonic force can achieve a throughput of 96 wells on a standard plate, and the limit state is 384 wells (suitable for phages with rapid growth and small morphology). In terms of throughput, it is 4 to 16 times that of the traditional method. Moreover, in the clinical application of phage therapy, it can quickly respond to clinical needs, timely feedback the phage infection effect on the targeted bacterial strain, and save about 10 times the time cost (taking 2 bacteria and 200 phages as an example, see the table below). On the premise of meeting accuracy, the present invention maximally saves manpower and material resources. In the field of basic scientific research, the present invention solves the problem of phage host spectrum determination, with high efficiency and high throughput. By accurately quantifying the area, radius, shape parameters, and halo size of plaques, it promotes in-depth research on the functions related to the phage's infectivity to the host, and provides an accurate characterization data basis for studying the interaction between phages and hosts. In summary, the high-throughput screening system for antibacterial phages is more accurate, time-saving, and cost-effective compared with the traditional double-layer agar plate method.

[0130] Method comparison Number of plates required * Manual operation time Number of spotting plates required * Manual operation time Total time consumed Traditional method 18 * 5 min 400*30s ～2h High-throughput rapid determination 2 * 5 min 2 min 12 min

[0131] A more optimized and faster image recognition algorithm. Due to the lack of plaque image data under standard parameters, in the future, on the premise of a large amount of data output, machine learning algorithms can be used to accelerate the upgrade and optimization of the phage-specific image recognition algorithm.

[0132] In the present invention, for the high-throughput screening system for antibacterial phages of the present invention, the spotting unit, the image acquisition unit, and the image recognition unit can be integrated in one device. For example, the spotting unit includes a mechanical device for spotting, such as using mechanical force or ultra-micro ultrasonic force; the image acquisition unit includes a photographing device; the image recognition unit is integrated in the computing unit of the device and is implemented by software and hardware. The above-mentioned image processing module and data analysis and visualization can also be integrated in the system. For example, a machine learning algorithm program is integrated in the computing unit of the device; visualization can be achieved through the display device configured in the system.

[0133] Example

[0134] Example 1 High-throughput determination of phage infection of host bacteria based on automated mechanical force and image enhancement recognition

[0135] Using mechanical implementation force (automated pipetting platforms such as Hamilton and multi-channel pipetting devices), prepare and pour double-layer agar plates containing host bacteria (such as aerobic strains like Acinetobacter baumannii, Klebsiella pneumoniae, Escherichia coli, etc.), and constantly pipette or dip the phage liquid to be detected for the host spectrum (usually 96-well throughput) onto the double-layer agar plates. Let the double-layer agar plates dry, and culture them under suitable growth conditions such as temperature, time, gas component concentration, etc.; finally, under camera parameters such as stable light source and focal length, conduct unified image shooting to obtain relevant image data. Through a system for high-throughput screening of antibacterial phages, different phages are distinguished by different plate coordinate positions. Each coordinate represents a phage to be detected, and each plate represents a host bacterium. Finally, measure the plaque area, radius, and transmittance (partial results are shown in Table 1), form a comprehensive score to describe the phage's ability to infect the host, obtain the host infection range of the target phage, and visualize the plaque score data using a software program. For example: For the infection determination of 64 phages and 125 Klebsiella pneumoniae strains, a total of 8125 image recognitions were carried out. The number of positive plaques with a comprehensive score > 1 was 864, and the positive rate was 864 / 8125 = 10.6%. Taking the bacterium "4049" as an example, among the 64 phages, 10 phages were able to infect it. The specific images are shown in Figure 6 , and the characterization scores and comprehensive scores are shown in Table 1. Among them, the comprehensive scores > 6 are the three phages represented by c2, d4, and g4. Therefore, compared with other phages, these three phages are strong phages that can infect the bacterium "4049". The same principle applies to the other 124 Klebsiella pneumoniae bacteria, and strong infecting phages can be screened out.

[0136] Table 1: Partial phage plaque quantitative analysis characteristics of infecting Klebsiella pneumoniae "4049" based on automated mechanical implementation force

[0137] Plate coordinate position Plaque area Plaque radius Transmittance Comprehensive score c1 934 17.24243 0.797297 4.886098 f1 3571 33.71475 0.748387 5.823371

[0138] h1 801.5 15.97264 0.814189 4.556266 c2 3010 30.9534 0.722581 6.403097 d4 3185.5 31.84299 0.717241 6.499633 g4 4219 36.64627 0.727891 6.263254 b9 4842 39.25884 0.753247 5.782442 g9 3092 31.37219 0.768212 5.486736 a10 5937.5 43.47373 0.767742 5.467716 h10 3576.5 33.74071 0.794702 4.945454

[0139] Example 2 High-throughput phage infection of host bacteria determination based on ultra-micro sound wave implementation force and image enhancement recognition

[0140] Apply force using ultrasonic waves (ultra-micro pipetting device Echo550), and constantly pipette the phage liquid to be detected of the host spectrum (which can be placed in 96-well, 384-well plates, etc.) onto a double-layer agar plate containing host bacteria (such as aerobic strains like Acinetobacter baumannii, Klebsiella pneumoniae, Escherichia coli, etc.). Air-dry the double-layer agar plate and culture it under suitable growth conditions such as appropriate temperature; finally, under stable light source, focal length and other camera parameters, conduct unified image shooting to obtain relevant image data. Finally, measure the plaque area, radius, and transmittance, form a comprehensive score to describe the phage's ability to infect the host, obtain the host infection range of the target phage, and visualize the plaque score data using a software program. For example: Conduct infection assays on 78 small phages and 125 Klebsiella pneumoniae strains, with a total of 9750 image recognitions. Among them, the number of positive plaques with a comprehensive score > 1 is 1055, and the positive rate is 1055 / 9750 = 10.8%. Taking Klebsiella pneumoniae "9384" as an example, among the 78 phages, 14 phages can infect it. The specific images are shown in Figure 7 , and the characterization scores and comprehensive scores are partially shown in Table 2. Among them, the two phages represented by b4 and c4 have a comprehensive score > 5. Therefore, compared with other phages, these two phages are strong phages that can infect bacteria "9384". Similarly, for the other 124 Klebsiella pneumoniae bacteria, strong infecting phages can be screened out. At the same time, conduct infection assays on 66 candidate phages and 58 Acinetobacter baumannii strains, with a total of 3828 image recognitions. Among them, the number of positive plaques with a comprehensive score > 1 is 949, and the positive rate is 949 / 3828 = 24.8%. Taking Acinetobacter baumannii "1050" as an example, among the 66 phages, 43 phages can infect it. The specific images are shown in Figure 8 , and the characterization scores and comprehensive scores are partially shown in Table 3. Among them, the four phages represented by b4, c4, b5, and d5 have a comprehensive score > 9. Compared with other phages, these four phages are the strongest phages that can infect bacteria "1050". Among them, the phages represented by c7 and f10 have a comprehensive score < 1, and after verification, they are negative plaques. Similarly, for the other 57 Klebsiella pneumoniae bacteria, the strongest infecting phages can be screened out.

[0141] Table 2: Partial phage plaque quantitative analysis characteristic data of phages that can infect Klebsiella pneumoniae "9384" based on ultra-micro sound wave applied force

[0142] Plate coordinate position Plaque area Plaque radius Transmittance Comprehensive score b1 417.5 11.52798 0.907975 2.725092 d2 575.5 13.53467 0.85625 3.732404 b3 673.5 14.64178 0.830189 4.223365 e3 613 13.96868 0.852564 3.783119 b4 712 15.05446 0.764331 5.615339 c4 722.5 15.16505 0.782051 5.193161 h4 571 13.48165 0.851613 3.78929 b5 346 10.49453 0.954839 1.691164 b6 563.5 13.39282 0.845161 3.915284 a7 646 14.33974 0.816456 4.434402 b7 376 10.94004 0.909677 2.69037

[0143] f7 448 11.94164 0.843137 4.000219 g7 632.5 14.18912 0.820261 4.425065 f10 275.5 9.364527 0.975155 1.134298

[0144] Table 3: Partial phage plaque quantitative analysis characteristic data of phages that can infect Acinetobacter baumannii "1050" based on ultra-micro sound wave applied force

[0145] Plate coordinate position Plaque area Plaque radius Transmittance Comprehensive score b1 1995.5 25.20292 0.670968 7.397241 d1 1593 22.51816 0.666667 7.480323 e1 3217.5 32.00253 0.653595 7.750551 f1 1239.5 19.86316 0.677632 7.23135 b2 1600.5 22.57111 0.647059 7.930215 f2 1188 19.44613 0.635762 8.1521 h2 2119 25.97111 0.671141 7.209961 a3 2309 27.11047 0.655844 7.612296 b3 2053.5 25.56657 0.606667 8.681814 c3 1397.5 21.09118 0.613333 8.623367 d3 1581 22.43319 0.610738 8.559694 e3 1756 23.64217 0.604027 8.75856 f3 1715 23.36453 0.610738 8.57825 h3 1244.5 19.90318 0.601351 8.846771 a4 617 14.01418 0.813333 4.609265 b4 2340 27.29185 0.581081 9.100288 c4 1335.5 20.61802 0.587838 9.148931 d4 2922.5 30.50017 0.619048 8.493838 e4 1368.5 20.8712 0.594595 8.988252 f4 1733 23.48683 0.615646 8.5375 h4 1337.5 20.63346 0.60274 8.802943 b5 1672 23.06977 0.585034 9.175529 c5 2987 30.83491 0.612245 8.644109 d5 1562 22.29798 0.591837 9.045079 e5 2924.5 30.51061 0.613793 8.41348 f5 3041 31.11238 0.616438 8.440125 d6 1950 24.91394 0.60274 8.764005 f6 1918.5 24.71189 0.6 8.815641 a7 1693.5 23.21762 0.647059 7.912119 b7 2530 28.37823 0.605442 8.770305 c7 186 7.69452 0.993103 0.747138

[0146] d7 2462 27.99427 0.630137 8.033872 f7 2222 26.59482 0.631944 7.954433 d8 1635.5 22.81657 0.594595 8.85463 f8 2096.5 25.83286 0.62069 8.465499 a9 1585.5 22.46509 0.655063 7.721857 d9 1374.5 20.91691 0.68 7.296408 h9 654 14.42826 0.979021 1.193435 a10 1514 21.9527 0.675 7.392247 b10 1488 21.76339 0.681818 6.936602 c10 1440.5 21.41321 0.677632 7.303152 e10 1667.5 23.0387 0.675676 7.323042 f10 278 9.40692 0.986301 0.724237

[0147] Example 3: High-efficiency Selection of Phage Monoclonal Guided by Image Recognition

[0148] The image recognition algorithm of the present invention can be integrated into a similar PIXL automated monoclonal screening device to add the function of selecting phage monoclonal. First, the image recognition unit of the present invention can accurately identify and locate the position of the plaque on the plate. Compared with the method without the image enhancement recognition algorithm (see Figure 9 A), Figure 9 As shown in B, the image recognition unit of the present invention can identify 22 plaque areas to be discriminated, Figure 9 while only 18 plaque areas to be discriminated can be identified as shown in A, with 4 missed screens. By repeatedly sampling and verifying a sample set with known positive infection relationships 3 times, the image recognition accuracy of this system is increased by 19%, and the positive missed screening rate is 1‰. Due to the high-precision image recognition, the high-efficiency selection of phage monoclonal can be realized. Using a mechanical device to drive the inoculation line, gun head, etc., the selection of phage monoclonal is realized under the guidance of laser. The inoculation process is completed by contacting the central area of the plaque with the inoculation line and gun head at a depth of 0.1 mm, and then it is moved to a liquid culture tube or a bacterial culture plate through a mechanical device to complete the inoculation and re-proliferation process, and finally phage monoclonal is obtained.

[0149] Example 4: High-throughput Screening and Verification of a Combined Phage Cocktail Preparation Applied to Clinical Phage Therapy

[0150] In the face of the clinical need for acute or chronic infections of antibiotic-resistant bacteria, using the existing phage library, high-throughput screening is carried out to screen for effective phages that can infect the target strain. Taking 175 phages targeting Acinetobacter baumannii as an example, the present invention can complete the pre-experiment for measuring the phage-host infection ability within 2 h. After the strain itself has grown for 16 hours, quantitative plaque characterization data can be obtained in a short time using the image enhancement recognition system to complete the verification of phage infection-related functions, and a script program is integrated and used for visual analysis of the phage infection ability (see Figure 4 ) Figure 4The darker the displayed color, the relatively stronger the infectivity. Thus, three highly infective phages can be screened out, with comprehensive scores all >6, as shown in Table 4. The entire screening process can be completed within 24 hours. Subsequently, these three phages are used to prepare a phage cocktail preparation. After liquid amplification culture, the titer is measured and diluted to 10^9 pfu / ml. They are combined in pairs to form experimental groups (X1: PH-AB1054B + PH-AB1052, X2: PH-AB1052 + PH-AB9078B, X3: PH-AB1054B + PH-AB9078B). After filtering endotoxins and other substances in the GMP workshop and preparing the preparation with each phage in a 1:1 ratio, this process takes 18 - 24 hours. Based on the ultrasonic implementation force, the bactericidal effect of the preparation is measured and quantitatively analyzed. Three groups are repeated and the average value is taken. The results are shown in Table 5, and this process takes 24 hours. From the analysis results, it can be obtained that the X3 preparation has a better killing effect on the bacterium "Ab333". The combination preparation is significantly superior to a single phage in terms of the transparency (i.e., transmittance) of the phage plaque, improving the infection efficiency. At the same time, to further verify the effect of the X3 preparation, after mixing the phage X3 preparation with the bacterial culture solution at different time points (0.5 h, 1 h, 1.5 h, 2 h), the liquid killing effect is observed, and 200 μl is taken for plating. The results are as Figure 10 shown, with almost no bacterial growth, further proving the bactericidal effect of the X3 preparation against the target host bacterium. In terms of functionality, it can be applied to clinical phage therapy.

[0151] Table 4: Three effective phages that can infect the strain "Ab333" by high-throughput screening

[0152] Bacteria ID Phage ID Plaque area Plaque radius Transmittance Comprehensive score

[0153] Ab333 PH-AB1054B 2349.0 27.34428500885924 0.7272727272727273 6.229744560610881 Ab333 PH-AB1052 2187.5 26.3875515352797 0.7315436241610739 6.240825882143517 Ab333 PH-AB9078B 2242.0 26.71424273349441 0.7304964539007093 6.258026081927733

[0154] Table 5: Verification of the infection effects of phage preparations X1, X2, and X3

[0155] Bacteria ID Phage preparation Plaque area Plaque radius Transmittance Comprehensive score Ab333 X1 2059.0 25.60078232500767 0.6906474820143885 7.063449052874526 Ab333 X2 2215.5 26.55589487929541 0.6861313868613139 7.156826280083204 Ab333 X3 2228.0 26.63070457981699 0.6788321167883211 7.268922457367736

[0156] Example 5 Verification of the specific functions of natural or synthetically modified phages

[0157] Aim at verifying the specific functions of phage's own genes, such as the gene swapping of tail fiber / capsid genes related to infection, gene knockout, etc. The technology of the present invention can be used to achieve high-throughput verification of the activity of virus vectors. Taking the modification of the dspB gene as an example, the natural unmodified phage PHKP4049 and the modified phage phkp4049dspB targeting specific genes are used. The mechanical pipetting semi-solid culture plate is used to grow for 20 h under suitable growth conditions. The unified parameters are used for image acquisition, and the phage function verification and quantitative analysis are completed by using the enhanced image algorithm. Then the data in Table 6 are obtained. In the comparison of four strains of bacteria, it is found that there is no change in the infection ability before and after modification in the infection performance of bacteria "4049" and "4023"; in bacteria "6360", it is found that the infection ability of the modified phage phkp4049dspB is limitedly enhanced, especially in the comparison of transmittance, with a difference of about 0.05, indicating that the phage plaque is clearer and more translucent, proving that the modification of the dspB gene affects the phage infection effect and is also affected by the host bacteria itself; at the same time, compared with PHKP4049, the radius of the phage plaque of phkp4049dspB is reduced by 8.044 units, indicating that the modification of the dspB gene may affect the phage replication ability.

[0158] Table 6: Infection characterization data of the natural unmodified phage PHKP4049 and the modified phage phkp4049dspB targeting four strains of bacteria

[0159] Bacteria ID Phage ID Plaque area Plaque radius Transmittance Comprehensive score 4023 phkp4049dspB 2229.5 26.63966762643186 0.7350993377483444 6.168146170323069 4023 PHKP4049 5706.5 42.61965937812973 0.7358490566037735 6.1405443689431 4049 phkp4049dspB 3092.0 31.37218781150401 0.7682119205298014 5.486736118763458 4049 PHKP4049 5937.5 43.4737271143878 0.7677419354838709 5.467716277433238 6360 phkp4049dspB 3949.0 35.45427675950801 0.7254901960784313 6.341204465071266 6360 PHKP4049 8991.5 53.49844242238791 0.7777777777777778 5.075320922625663

[0160] Example 6 Parallel comparison with the prior art

[0161] To intuitively compare the utility of the present invention, the inventors respectively conducted parallel comparisons of the one-step growth curve of phage and the bactericidal curve of phage. In the traditional conventional research of phage, the experimental curve quantitatively describing the growth law of virulent phage is called the one-step growth curve. Taking the natural phage PHAB8962 as the experimental sample, the one-step growth curve results plotted by measuring the PFU at each time point within 2 hours of the phage are as Figure 11 shown. The infection characterization data of the phage PHAB8962 are shown in Table 7. By comparison, it can be obtained that the one-step growth curve does not intuitively describe the phage infection ability. The bactericidal curve of phage is often regarded as the gold standard for the phage's ability to infect bacteria and is generally recognized as more accurate. Therefore, the present invention conducts a parallel comparison for it. The bactericidal curve is measured with the natural phages PHKP4049 and PHKP9822-B1 as the test objects. The host bacteria Kp9816 are inoculated into the lysogeny broth medium (LB) and cultured until OD600 = 1.7. 150 μl of the bacterial solution is added to each well of the 96-well plate, and then the phage is added to make their MOI (the ratio of the number of phages to the number of indicator bacteria cells during infection is called the multiplicity of infection) be 0.01 (Figure 12 A) and 0.05( Figure 12 B). At this time, the PFU of phage PHKP4049 was 2.64×10^9, and the PFU of phage PHKP9822-B1 was 1.6×10^8. The infection results were recorded after 20 h using an Epoch2 microplate reader. The blank control was a blank culture medium, and the bactericidal curve was obtained as Figure 12 shown. Compared with the control Kp9816 group (bacterial solution without phage added), both phages had bactericidal ability. The infection characterization data of phages PHKP4049 and PHKP9822-B1 are shown in Table 8. Using the infection ability score obtained by the method of the present invention, phage PHKP4049 was higher than phage PHKP9822-B1. Compared with Figure 12 the bactericidal curve results, the results of the two methods were consistent. The infection and bactericidal ability of phage PHKP4049 was higher than that of phage PHKP9822-B1, which confirmed the accuracy of the method of the present invention.

[0162] Three methods were evaluated from three aspects. In terms of cost: for the one-step growth curve method, 39 culture plates were required for measuring 13 time points to calculate the phage PFU, and it took 24 h; one 96-well plate was required for the bactericidal curve, and it usually took 24 - 25 h; the present invention only required one square culture plate and took 16 - 24 h. In terms of accuracy: the method of the present invention was compared with the other two methods respectively, which proved that phage PHAB8962 could infect bacteria "AB8962", and the reliable result that the infection and bactericidal ability of phage PHKP4049 was higher than that of phage PHKP9822-B1, thus indicating the accuracy of the method of the present invention. At the same time, it can be obtained that the one-step growth curve was greatly affected by the operator's own operation and relied on manual measurement; the bactericidal curve reflected the phage characteristics through the OD value of bacteria, was easily affected by the experimental environment, and could obtain a relatively accurate qualitative result (i.e., whether it could infect), but the quantitative ability was insufficient. In terms of application and promotion: the one-step growth curve only represented the number of phages in each growth period and mainly reflected the growth characteristics; the bactericidal curve needed multiple groups of controls to obtain results and was often used for exploring a single influencing variable; the present invention could achieve data output at different levels through systematic quantification, and the results were more intuitive and easier to be popularized and applied in the phage field.

[0163] In summary, compared with the traditional methods, the present invention was not less accurate. The advantages were the reduction of labor costs, the ability to achieve high throughput through different implementation forces, and a wider application scenario through quantitative analysis of image recognition data.

[0164] Table 7: Plaque characterization data of phage PHAB8962

[0165] Bacteria ID Phage ID Plaque area Plaque radius Transmittance Comprehensive score AB8962 PHAB8962 617.0 14.01417852659937 0.813333333333333 4.609265168091537

[0166] Table 8: Plaque characterization data of phage PHAB8962

[0167] Bacteria ID Phage ID Plaque area Plaque radius Transmittance Comprehensive score Kp9816 PHKP4049 641.5 14.28971 0.811594 4.624668 Kp9816 PHKP9822-B1 396.5 11.23432 0.931298 2.196365

Claims

1. A method for high-throughput screening of antibacterial phages, the method comprising: 1) Spotting a plurality of phages to be detected on a plate containing host bacteria and culturing; 2) Taking an image of the cultured plate to obtain a plaque image; 3) Binarizing the plaque image to convert the phage image into a binary image; 4) Filtering the binary image based on the shape and transmittance of the plaques in the binary image to obtain plaque antibacterial phenotype data; 5) Screening the phages to be detected based on the phage antibacterial phenotype data to obtain the antibacterial phages.

2. The method according to claim 1, in 1), the plate is a double-layer agar plate.

3. The method according to claim 2, in 1), the plurality of phages to be detected are spotted in an array on a double-layer agar plate containing host bacteria.

4. The method according to any one of claims 1-3, in 1), using a mechanical implementation force or an ultrasonic implementation force, sucking or dipping a phage liquid of a plurality of host spectra to be detected and spotting it on a plate containing host bacteria.

5. The method according to any one of claims 1-3, in 2), enhancing the plaque image to eliminate the influence of uneven illumination on the image.

6. The method according to claim 5, using the Retinex algorithm to enhance the plaque image.

7. The method according to claim 5, enhancing the plaque image by the following steps: (a) The obtained original image S(x, y) is expressed as the product of a illumination image L(x, y) and a reflectance image R(x, y), as shown in Equation (1): S(x, y) = R(x, y)*L(x, y) (1); Taking the logarithm of both sides of Equation (1) gives: log[R(x, y)] = log[S(x, y)] – log[L(x, y)] (2); (b) Approximately obtaining the illumination image L(x, y) by convolving the original image S(x, y) with a Gaussian kernel G(x, y), and converting Equation (2) gives Equation (3); log[R(x, y)] = log[S(x, y)] – log[S(x, y)*G(x, y)] (3); (c) Quantifying the obtained log[R(x, y)] result into pixel values in the range of [0, 255], and then outputting the result image R(x, y): R(x, y) = (log[R(x, y)] – Min) / (Max – Min)*255 (4); Among them, Min is the minimum value in log[R(x, y)], and Max is the maximum value in log[R(x, y)].

8. The method according to any one of claims 1-3, 6, 7, in 3), binarizing the plaque image by fusing two sizes.

9. The method according to any one of claims 1-3, 6, 7, in 4), the shape of the plaque in the binary image refers to the circularity of the plaque, which is calculated by the following formula (5): Circularity = 4×π×Area÷(Perimeter×Perimeter) (5), Among them, The area and perimeter are respectively the area and perimeter of the plaque in the binary image.

10. The method according to claim 9, wherein the filtering is to remove plaques with a circularity lower than 0.

1.

11. The method according to any one of claims 1-3, 6, 7, 10, in 4), the transmittance of the plaque in the binary image is calculated by the following formula (6), Transmittance of plaque = Plaque brightness value / Background brightness value (6).

12. The method according to claim 11, wherein the filtering is to remove plaques with a circularity lower than 0.

1.

13. According to the method described in claim 11, in 5), the comprehensive score is calculated using the shape and transmittance of the plaque, the plaque is judged according to the comprehensive score, and the phage to be detected is screened based on the plaque judgment result to obtain the antibacterial phage, wherein, The comprehensive score is calculated by the following formula (7): Comprehensive score = Circularity + 20×(1 - Transmittance of plaque) (7).

14. The method according to claim 13, plaques with a comprehensive score greater than 1 are judged as positive plaques, and the phages to be detected corresponding to the positive plaques are antibacterial phages.

15. A high-throughput screening system for antibacterial phages, the system comprising: A phage culture unit for spotting multiple phages to be detected on a plate containing host bacteria and culturing them; An image acquisition unit for photographing the cultured plate to obtain a plaque image; An image binary conversion unit for converting the phage image into a binary image; A data extraction unit for filtering the binary image based on the shape and transmittance of the plaques in the binary image to obtain plaque antibacterial phenotype data; An analysis unit for screening the phages to be detected based on the plaque antibacterial phenotype data to obtain the antibacterial phages.

16. The system according to claim 15, the system further comprising: An image enhancement unit for enhancing the plaque image and eliminating the influence of uneven illumination on the image.

17. The system according to claim 15 or 16, in the phage culture unit, the multiple phages to be detected are spotted on a plate containing host bacteria in an array.

18. The system according to claim 15 or 16, in the phage culture unit, using a mechanical force or an ultrasonic force, sucking or dipping the phage liquid of multiple host spectra to be detected and spotting it on a plate containing host bacteria.

19. The system according to claim 15 or 16, in the image enhancement unit, the Retinex algorithm is used to enhance the plaque image.

20. The system according to claim 19, the following steps are used to enhance the plaque image: (a) The acquired original image S(x, y) is expressed as the product of the illumination image L(x, y) and the reflectance image R(x, y), as shown in formula (1): S(x, y) = R(x, y)*L(x, y) (1); Taking the logarithm of both sides of formula (1) gives: log[R(x, y)] = log[S(x, y)] – log[L(x, y)] (2); (b) The illumination image L(x, y) is approximately obtained by convolving the original image S(x, y) with a Gaussian kernel G(x, y). Equation (3) can be obtained by transforming Equation (2); log[R(x, y)] = log[S(x, y)] – log[S(x, y)*G(x, y)] (3); (c) The obtained result of log[R(x, y)] is quantized into pixel values in the range of [0, 255], and then the result image R(x, y) is output, as shown in Equation 4: R(x, y) = (log[R(x, y)] – Min) / (Max – Min)*255 (4); Among them, Min is the minimum value in log[R(x, y)], and Max is the maximum value in log[R(x, y)].

21. The system according to any one of claims 15, 16, and 20, in the image binary conversion unit, two sizes are fused to perform binaryzation on the plaque image.

22. The system according to any one of claims 15, 16, and 20, in the data extraction unit, the shape of the plaque in the binary image refers to the circularity of the plaque, which is calculated by the following formula (5): Circularity = 4×π×Area÷(Perimeter×Perimeter) (5), Among them, The area and perimeter are respectively the area and perimeter of the plaque in the binary image.

23. The system according to claim 22, the filtering is to remove plaques with a circularity lower than 0.

1.

24. The system according to any one of claims 15, 16, 20, and 23, in the data extraction unit, the transmittance of the plaque in the binary image is calculated by the following formula (6), Transmittance of the plaque = Brightness value of the plaque / Brightness value of the background (6).

25. The system according to claim 24, the filtering is to remove plaques with a transmittance greater than 0.

995.

26. The system according to any one of claims 15, 16, 20, 23, and 25, in the data extraction unit, calculates a comprehensive score using the shape and transmittance of the plaque, makes a plaque judgment based on the comprehensive score, and screens the phage to be detected based on the plaque judgment result to obtain the antibacterial phage, wherein, The comprehensive score is calculated by the following formula (7): Comprehensive score = Circularity + 20×(1 - Transmittance of the plaque) (7).

27. The system according to claim 26, plaques with a comprehensive score greater than 1 are judged as positive plaques, and the phages to be detected corresponding to the positive plaques are antibacterial phages.

28. A method for identifying phage infectivity, selecting phage monoclonal, formulating a phage cocktail preparation for phage therapy, verifying phage function, reverse screening for phage-resistant host bacteria or verifying phage lyase function, the method includes screening antibacterial phages using the method according to any one of claims 1-14.

29. The method according to claim 28, the method for formulating a phage cocktail preparation for phage therapy includes: 1) Screening effective phages that target and infect the target bacteria and quantitatively analyzing antibacterial infection characteristics, and screening candidate phages according to the phage cocktail preparation standard; 2) Amplifying and culturing the candidate phages and formulating a phage cocktail preparation; 3) Evaluating the effect of the phage cocktail preparation on the target bacteria.

30. The method according to claim 29, wherein the cocktail preparation standard comprises: 1) It is only effectively bactericidal against the target bacteria and not effectively bactericidal against other bacteria; 2) It can effectively exert its effect and effectively reduce the off-target effect; 3) The host spectrum ranges of different phages can be complementary.

31. The method according to claim 29 or 30, wherein steps 1) and 3) are carried out using the system according to any one of claims 15-27.

Images