Matched data management system of chicken feed utilization rate character molecular detection kit

Abstract

The invention discloses a chicken feed utilization rate character molecular detection kit matched data management system, particularly relates to the field of poultry genetic detection, and aims to solve the problem of feed utilization rate typing result deviation. The spatial drift and the time drift are synchronously stripped at a signal source; a curve generated after real-time correction is subjected to global screening through a machine learning model, abnormal signals are isolated immediately in high-throughput operation, and amplification features are strictly aligned with physical positions; a subsequent allele map and a feeding index are integrated at a time on a data management platform, and each feed conversion track has a unique and credible genetic fingerprint; the thermal light field model is self-learned along with batches, the drift compensation precision keeps convergence along with the detection quantity, and the drift compensation precision is not dragged by hidden deviation after long-term operation; a breeding window is compressed from a periodic level to near real-time, the feed utilization rate improvement rhythm is accelerated, and resource investment is more controllable.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of poultry genetic detection, more particularly, the present application relates to a chicken feed utilization trait molecular detection kit matching data management system. BACKGROUND

[0002] The molecular detection center uses a closed fluorescence genotyping pipeline to process the feed utilization trait detection kit. Each kit is pre-packaged with primers, probes and stabilizers. After the reaction solution is dispensed, it is loaded into a ring-shaped heating channel. The optical module captures the fluorescence curve and immediately pushes the data to the feed utilization data management platform. The platform automatically calls the sample barcode and establishes a one-to-one mapping with the feeding record and weight gain log, aiming to restore the precise correlation between genotype and production performance in real time. The whole process is highly automated with minimal human intervention. The physical properties of the equipment and algorithm interpretation play a key role in the data link.

[0003] The thermal radiation in the closed channel forms a subtle thermal gradient at the edge of the reagent plate, and the light path refraction difference further raises the edge hole background fluorescence. The interpretation program uses a unified threshold strategy and does not identify the difference between hole positions, thus misjudging the slight drift as a real amplification signal. The false genotype is written into the platform along with the detection results and matched with the feeding curve in real time. Due to fluctuations in the feeding record itself, the misjudgment signal and noise are superimposed on each other and difficult to expose in regular quality control, eventually raising the feed conversion evaluation in the genetic evaluation link, deviating from the breeding target and making it difficult to trace the source.

[0004] To solve the above problems, a technical solution is provided. SUMMARY

[0005] To overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a chicken feed utilization trait molecular detection kit matching data management system. By freezing the full-plate light and temperature grid at the moment of sample addition and recursively updating the thermal light field calibration template, spatial drift and temporal drift are simultaneously stripped at the signal source. The generated curve after real-time correction is then screened by a machine learning model, and abnormal signals are immediately isolated in high-throughput operation. The amplification characteristics and physical positions are strictly aligned. The subsequent allele map and feeding index are integrated on the data management platform at one time, and each feed conversion track has a unique and reliable genetic fingerprint. The thermal light field model is self-learning with each batch, and the drift compensation accuracy converges with the detection amount. Long-term operation is no longer dragged by hidden bias. The breeding window is compressed from a cycle to near real-time, and the feed utilization improvement pace is accelerated and the resource investment is more controllable, thus solving the problems raised in the background art.

[0006] To achieve the above-mentioned purposes, the present application provides the following technical solutions: Light unit: At the end of sample loading, the initial fluorescence intensity and temperature of all wells on the reagent plate are synchronously collected, and the coordinates are recorded to generate the initial position-light temperature grid. Pre-drift unit: Based on the initial position-light temperature grid, the convolution interpolation operation is performed to output the well drift prediction surface and the corresponding drift mask, and to distinguish the edge group and the center group background. Calibration unit: The drift mask is weighted and superimposed on the real-time amplification curve to obtain the drift correction curve set after removing the background drift component. Spectrum evaluation unit: Extract multi-dimensional key features from the drift correction curve set, input the historical data pre-trained machine learning model to evaluate the amplification integrity, and generate the allele spectrum or mark abnormal samples according to the evaluation results. Abnormal storage unit: Write the allele spectrum corresponding to the feeding mapping index into the data management platform, and list the abnormal sample number in the review queue. More module unit: Write the well drift prediction surface and its corresponding correction parameters into the next batch initialization table to be used as a recursive thermal light field calibration template.

[0007] In a preferred embodiment, the light unit operates as follows: After the reaction liquid sample is loaded, the optical module and temperature sensor are immediately started to synchronously scan all wells on the reagent plate, record the initial fluorescence intensity and initial temperature of each well, assign a unique two-dimensional coordinate to each well on the reagent plate as its spatial position identifier, combine the initial fluorescence intensity and initial temperature of each well into a data pair, and organize it into a two-dimensional matrix according to the two-dimensional coordinate to generate the initial position-light temperature grid. Store the initial position-light temperature grid to the database to ensure that the data pair of each well is bound to its two-dimensional coordinate.

[0008] In a preferred embodiment, the pre-drift unit operates as follows: Based on the initial position-light temperature grid, perform interpolation operation on the fluorescence intensity and temperature data using a Gaussian convolution kernel to generate continuous fluorescence intensity prediction surface and temperature prediction surface. Extract the predicted value of each well from the fluorescence intensity prediction surface and the temperature prediction surface, calculate the fluorescence drift and the temperature drift, and generate the well drift prediction surface. Based on the well drift prediction surface, calculate the drift mask value of each well to identify the drift severity. Divide the wells into edge group and center group, calculate the average drift mask value of the edge group and the center group, and quantify the background drift difference.

[0009] In a preferred embodiment, the calibration unit operates as follows: The fluorescence intensity of each well site is captured immediately after each heating cycle by the optical module to obtain real-time amplification curves and generate time series data; a drift correction factor is calculated according to the period length of the heating cycle and the current cycle number, and the correction strength is dynamically adjusted by period weighting of the drift mask.

[0010] In a preferred embodiment, the drift correction unit operation further includes the following: The drift correction factor is applied to the original fluorescence intensity at each time point to perform drift correction of the real-time amplification curve and generate a corrected fluorescence intensity time series; the corrected curves of all well sites are integrated into a drift correction curve set to form a complete data set.

[0011] In a preferred embodiment, the spectrum evaluation unit operation includes the following: The corrected fluorescence intensity time series of each well site is extracted from the drift correction curve set; the second derivative of the corrected fluorescence intensity time series of each well site is calculated to identify a candidate inflection point set, and the inflection point with the largest curvature is selected as the main inflection point, and its curvature value is recorded as the inflection point clarity score; the area before the inflection point and the area after the inflection point are calculated with the main inflection point as the boundary, and then the signal persistence ratio is calculated; the inflection point clarity score and the signal persistence ratio of each well site are input into the support vector machine model pre-trained based on historical data, and the amplification completeness index is output.

[0012] In a preferred embodiment, the spectrum evaluation unit operation further includes the following: According to the comparison of the amplification completeness index with the preset threshold value, it is determined whether to extract the inflection point sequence from the corrected curve to generate an allele spectrum, or to mark the well site as an abnormal sample and record the number.

[0013] In a preferred embodiment, the abnormality storage unit operation includes the following: For each allele spectrum, the corresponding sample barcode is searched in the sample barcode set according to the well site coordinates corresponding to the allele spectrum to generate a corresponding record containing the allele spectrum and the sample barcode; for each corresponding record containing the allele spectrum and the sample barcode, the feeding data and weight gain data corresponding to the sample barcode are extracted by searching the feeding mapping index to generate a complete data combination containing the allele spectrum, the feeding data and the weight gain data; for each well site identified as abnormal, the corresponding sample barcode is searched in the sample barcode set according to the well site coordinates to generate a review record containing the sample barcode and the well site coordinates, and the review record is integrated into a review queue.

[0014] In a preferred embodiment, the more module unit operation includes the following: The well position drift prediction surface of the current batch is stored to the database, and the fluorescence drift and temperature drift of each well position are recorded with the batch number as the index; for each well position, the average value of the drift correction factor at all time points of the current batch is calculated to obtain the average drift correction factor, which is stored together with the well position drift prediction surface.

[0015] In a preferred embodiment, the more module unit operation further includes the following: According to whether the current batch is the first batch, if the current batch is the first batch, the well position drift prediction surface and the average drift correction factor are directly used as the initialization table of the next batch, and if there is an initialization table of the previous batch, the well position drift prediction surface and the average drift correction factor of the previous batch are fused with the data of the current batch by weighted average to generate a recursive thermal light field calibration template for the next batch.

[0016] The technical effects and advantages of the chicken feed utilization trait molecular detection kit supporting data management system of the present application are as follows: By freezing the full-plate light and temperature grid at the moment of sample addition and recursively generating the thermal light field calibration template, spatial drift and temporal drift are simultaneously stripped at the signal source; the generated curve after real-time correction is further screened by a machine learning model, and abnormal signals are immediately isolated in high-throughput operation; the subsequent allele map and feeding index are integrated on the data management platform, and common missing and mismatching mapping by hand is fundamentally eliminated, so that each feed conversion track has a unique and reliable genetic fingerprint; the thermal light field model learns from batch to batch, and the drift compensation accuracy converges with the detection amount, so that long-term operation is no longer dragged by hidden bias; the entire detection-analysis-archiving-decision link is closed in a single data channel, and the selection window is compressed from a cycle to near real-time, so that the feed utilization improvement rhythm is accelerated and resource investment is more controllable. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The structural diagram of the chicken feed utilization trait molecular detection kit supporting data management system of the present application is shown in the figure. Figure 2 The structural diagram of the abnormality storage unit of the chicken feed utilization trait molecular detection kit supporting data management system of the present application is shown in the figure. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0019] Embodiment 1: Figure 1 The application discloses a chicken feed utilization trait molecular detection kit matching data management system, comprising: Lighting unit: at the end of sample addition, the initial fluorescence intensity and temperature of all wells on the reagent plate are synchronously collected, and the coordinates are recorded to generate an initial position-light temperature grid.

[0020] Pre-bleaching unit: based on the initial position-light temperature grid, convolution interpolation operation is performed to output well position drift prediction surface and corresponding drift mask, and to distinguish edge group and center group background.

[0021] Curve correction unit: the drift mask is weighted and superimposed on the real-time amplification curve, and after removing the background drift component, a drift corrected curve set is obtained.

[0022] Spectrum evaluation unit: multi-dimensional key features are extracted from the drift corrected curve set, and a machine learning model pre-trained with historical data is input to evaluate amplification completeness, and an allele spectrum or a marked abnormal sample is generated according to the evaluation result.

[0023] Abnormality storage unit: the allele spectrum is written into the data management platform after being one-to-one corresponding to the feeding mapping index, and the abnormal sample number is listed in the review queue.

[0024] More module unit: the well position drift prediction surface and the corresponding correction parameters are written into the next batch initialization table to be used as a recursive thermal light field calibration template.

[0025] In modern animal husbandry, the optimization of chicken feed utilization rate is not only the core of improving economic benefits of breeding, but also the key path to realize sustainable development. The chicken feed utilization trait molecular detection kit matching data management system uses high-throughput molecular detection technology combined with automatic data processing to accurately evaluate the feed utilization trait of chickens and provide scientific basis for breeding selection and feeding management. The application relies on a closed fluorescence typing pipeline, uses pre-packaged primers, probes and stabilizers, captures fluorescence signals through a ring-shaped heating channel and an optical module after reaction liquid dispensing, and pushes data to a data management platform in real time, which is associated with feeding records and weight gain logs. However, due to the deviation caused by physical factors (such as thermal gradient) and optical factors (such as light refraction difference) in the detection process, the fluorescence signal of the edge well may drift, which may affect the accuracy of genotype interpretation, and even accumulate errors in genetic evaluation, deviating from the breeding target. To solve this problem, the lighting unit, as the starting point of the whole process, collects initial data at the end of sample addition, lays a foundation for subsequent correction and analysis, and ensures the reliability of detection results and the accuracy of production data.

[0026] End of loading refers to the time when, in molecular detection experiments, especially when using techniques such as fluorescent quantitative polymerase chain reaction, the reaction solution (including primers, probes, enzymes, dNTPs, buffers, etc.) and the sample to be tested (such as DNA or RNA) are added to all reaction wells of the reagent plate, and the loading operation is completed. At this time, each reaction well on the reagent plate contains a complete reaction system, ready for the subsequent heating cycle stage, but the reaction has not officially started.

[0027] In the typical process of molecular detection technology, the experimental operation usually includes the following main links: Sample preparation: extract and purify nucleic acids (such as DNA or RNA) in the sample to be tested.

[0028] Reaction system preparation: mix the reaction solution components with the sample to be tested according to a certain proportion to prepare a complete reaction system.

[0029] Loading: The prepared reaction system is divided into each reaction well of the reagent plate.

[0030] Thermal cycling reaction: through heating denaturation, annealing and extension steps, nucleic acid amplification and real-time monitoring of fluorescence signals are carried out.

[0031] Data analysis: analyze the content or characteristics of the target nucleic acid in the sample according to the collected fluorescence signals.

[0032] End of loading is specifically located at the transition point between the third step (loading) and the fourth step (thermal cycling reaction) in the above process. In other words, it is the end of the loading operation and the preparation state before the start of the thermal cycling reaction. At this time, the reaction system has been completely distributed to each well of the reagent plate, but the heating cycle has not started, the reaction has not occurred, and the fluorescence signal in the well is mainly derived from background fluorescence and reagent autofluorescence, rather than the signal generated by the amplification reaction.

[0033] Selecting the moment of end of loading to collect data has specific technical significance. The fluorescence intensity and temperature data at this moment can be used as an initial reference for subsequent analysis and correction. Specifically: Initial fluorescence intensity: reflects the background fluorescence level of the well after loading, providing a reference for the change of fluorescence signal in the subsequent reaction process.

[0034] Initial temperature distribution: records the temperature state of the well after loading, facilitating the analysis of the influence of thermal gradient on fluorescence signal.

[0035] Data correction basis: based on the initial state at this time, the signal deviation caused by physical factors (such as optical path refraction difference or thermal drift) in the reaction process can be corrected.

[0036] Therefore, the end of loading is not only a time node in operation, but also a key starting point for data collection and analysis.

[0037] The purpose of the light unit is to capture the initial fluorescence intensity and temperature state of all wells on the reagent plate at the moment when the sample addition is completed, and to generate an initial position-light temperature grid as a reference for subsequent drift correction and data analysis. The following is a detailed description of the specific processing technology logic: Data collection: After the reaction liquid sample addition is completed, immediately start the optical module and temperature sensor to perform synchronous scanning on all wells on the reagent plate to record the initial state of each well. The optical module detects the initial fluorescence intensity of each well through optical signal detection, indicating the optical characteristics of the reaction system at the beginning of detection. The temperature sensor simultaneously measures the initial temperature of each well, reflecting the thermal state at the end of sample addition. The collection of fluorescence intensity and temperature data is strictly aligned in time to ensure that both describe the state of the well at the same time, avoiding data deviation due to time difference.

[0038] This synchronous collection design can truly reflect the state of the reaction system before the start of the heating cycle, providing accurate starting point data for subsequent drift correction. By ensuring the consistency of fluorescence intensity and temperature data, the physical and optical characteristics of the initial state are fully retained, facilitating accurate tracing of the starting conditions of the detection in subsequent analysis.

[0039] Well coordinate recording: Each well on the reagent plate is assigned a unique two-dimensional coordinate to identify its spatial position. The structure of the reagent plate usually contains multiple rows and columns, with row coordinates increasing from the first row to the last row, and column coordinates increasing from the first column to the last column. For example, a reagent plate containing 8 rows and 12 columns has a well coordinate range of row coordinates from 1 to 8 and column coordinates from 1 to 12. The coordinates of each well are composed of its row number and column number, forming a unique position identifier.

[0040] By setting fixed coordinates for each well, we can accurately locate and reference the data of a specific well throughout the entire data processing process, avoiding processing errors caused by position confusion. This coordinate recording method makes data management more orderly, ensuring spatial accuracy in subsequent steps when processing well-related data, and improving the accuracy and reliability of data reference.

[0041] Initial position-light temperature grid generation: The initial fluorescence intensity and initial temperature of each well position are combined into a data pair representing the initial state characteristics of the well position. Then, according to the two-dimensional coordinates of each well position, the data pairs of all well positions are organized into a two-dimensional matrix, named initial position-light temperature grid. The number of rows of the two-dimensional matrix corresponds to the number of rows of the reagent plate, and the number of columns corresponds to the number of columns of the reagent plate. Each element in the matrix is the data pair of the corresponding well position. For example, for a reagent plate with 8 rows and 12 columns, the initial position-light temperature grid is a matrix containing 8 rows and 12 columns. Each matrix element stores the initial fluorescence intensity and initial temperature of the position.

[0042] This data organization method can intuitively reflect the spatial distribution characteristics of the well positions on the reagent plate, and facilitate subsequent steps to use spatial information for drift prediction and correction processing. The initial position-light temperature grid not only retains the independent data of each well position, but also reflects the relative position relationship between the well positions through the matrix form, providing necessary structural support for subsequent spatial consistency analysis, thereby improving the efficiency and accuracy of data processing.

[0043] Data storage: The generated initial position-light temperature grid is stored in the database of the data management platform to ensure that the data pair of each well position is closely bound to its two-dimensional coordinates. When storing, the database records the row coordinates, column coordinates, and corresponding initial fluorescence intensity and initial temperature of each well position, forming a queryable data set. This binding method ensures the integrity and position correlation of the data, facilitating the subsequent steps to directly extract the data of a specific well position from the database.

[0044] The light collecting unit synchronously collects the initial fluorescence intensity and temperature of all well positions on the reagent plate at the moment when the sample addition is completed, and generates the initial position-light temperature grid in combination with the two-dimensional coordinates, providing accurate initial state description for the subsequent processes of the data management system for the chicken feed utilization trait molecular detection kit. This grid completely records the optical and thermal characteristics at the beginning of the detection, avoiding the risk of not capturing the initial deviation caused by environmental factors such as thermal gradient or optical path difference. The initial position-light temperature grid will be directly referenced by the subsequent steps for calculating well position drift prediction surface and drift mask, ensuring the spatial consistency and data traceability of the correction process, thereby improving the accuracy of feed utilization trait evaluation.

[0045] The light unit generates an initial position-light temperature grid by collecting a full-plate blank fluorescence snapshot at the end of sample addition and recording the well coordinates, providing baseline data for subsequent drift correction and data analysis. The initial position-light temperature grid records the initial state of fluorescence intensity and temperature of each well on the reagent plate, which is the key to capturing the spatial distribution characteristics of the reagent plate. However, due to the thermal radiation in the closed channel forming a subtle thermal gradient at the edge of the reagent plate, the background fluorescence of the edge wells is higher than that of the center wells due to the refraction difference of the light path, resulting in differences in the initial state between wells. If not handled, this difference will be misjudged as a real amplification signal, affecting the genetic evaluation accuracy of feed utilization traits. The pre-drift unit generates a drift prediction surface and a drift mask based on the initial position-light temperature grid through a specific spatial analysis method, distinguishing the background differences between the edge group and the center group, and providing accurate spatial correction basis for subsequent drift correction of real-time amplification curves.

[0046] The pre-drift unit takes the initial position-light temperature grid as input, generates a well drift prediction surface and a drift mask through a series of calculations and analyses, and distinguishes the background differences between the edge group and the center group. The following is a detailed description of the specific processing technology logic: Convolution interpolation operation: The initial position-light temperature grid contains initial fluorescence intensity and initial temperature data for each well, but these data only reflect discrete well position information and cannot directly show the continuous distribution trend of fluorescence intensity and temperature on the reagent plate. To solve this problem, the pre-drift unit first performs a convolution interpolation operation on the initial position-light temperature grid. The Gaussian convolution kernel is selected as the interpolation tool, and its parameters are determined according to the physical size and thermal gradient characteristics of the reagent plate to smooth the mutations of fluorescence intensity and temperature between wells. The specific processing process is as follows: based on the initial fluorescence intensity data, combined with the weight distribution of the Gaussian convolution kernel, a weighted average calculation is performed on the surrounding area of each well to generate a continuous fluorescence intensity prediction surface; similarly, based on the initial temperature data, the same Gaussian convolution kernel weight distribution is used to calculate and generate a continuous temperature prediction surface. The fluorescence intensity prediction surface and the temperature prediction surface respectively reflect the spatial distribution trend of fluorescence intensity and temperature on the reagent plate, providing complete spatial information support for subsequent analysis.

[0047] Well drift prediction surface generation: The fluorescence intensity prediction surface and the temperature prediction surface generated based on the convolution interpolation operation are further used to calculate the drift amount of each well position to quantify the change of the well position relative to the initial state. The processing process is as follows: for each well position on the reagent plate, the predicted value corresponding to the position of the well position is extracted from the fluorescence intensity prediction surface, and the initial fluorescence intensity value is subtracted to obtain the fluorescence drift amount; similarly, the predicted value corresponding to the position of the well position is extracted from the temperature prediction surface, and the initial temperature value is subtracted to obtain the temperature drift amount. The fluorescence drift amount and the temperature drift amount respectively represent the change degree of each well position in fluorescence intensity and temperature. The fluorescence drift amount and the temperature drift amount of all well positions are organized according to the well position to form a well position drift prediction surface. The well position drift prediction surface clearly shows the drift distribution of each well position on the reagent plate, and provides necessary data support for the generation of the subsequent drift mask.

[0048] Drift mask generation: In order to evaluate the drift severity of each well position, the pre-drift unit calculates a drift mask based on the well position drift prediction surface to provide an intuitive drift quantification index. The specific processing process is as follows: for each well position, the absolute value of the fluorescence drift amount is taken, and divided by the maximum value in the absolute values of the fluorescence drift amounts of all well positions to obtain the normalization result of the fluorescence drift amount; similarly, the absolute value of the temperature drift amount is taken, and divided by the maximum value in the absolute values of the temperature drift amounts of all well positions to obtain the normalization result of the temperature drift amount. Then, the fluorescence drift amount normalization result of each well position is added to the temperature drift amount normalization result to obtain the drift mask value. The range of the drift mask value is controlled between 0 and 2, and the larger the value is, the more significant the drift degree of the well position is. The generation process of the drift mask eliminates the influence of the dimension difference between the fluorescence and temperature data through normalization processing, making the drift degree evaluation more consistent and comparable, and providing a reliable basis for the correction of the real-time amplification curve.

[0049] Distinguishing edge group and center group background: The edge well positions of the reagent plate are greatly affected by the thermal gradient and light path refraction, and there are obvious differences in the drift characteristics with the center well positions, so the well positions are divided into an edge group and a center group, and the background drift characteristics of the two groups are calculated respectively. The edge group is defined as the well positions in the outermost circle of the reagent plate, and the center group is defined as the well positions inside the reagent plate. The processing process is as follows: for all well positions of the edge group, the average value of the drift mask values thereof is calculated to obtain the average drift mask value of the edge group; for all well positions of the center group, the average value of the drift mask values thereof is calculated to obtain the average drift mask value of the center group. By comparing the average drift mask value of the edge group and the average drift mask value of the center group, the difference in the background drift of the well positions in the two groups is quantified. This grouping calculation method can reflect the drift characteristics of different regions of the reagent plate, and provide regionalized background information support for subsequent analysis and correction.

[0050] The pre-drift unit generates continuous fluorescence intensity prediction surface and temperature prediction surface by performing convolution interpolation operation on the initial position-light temperature grid, and then calculates the hole position drift prediction surface and drift mask, and distinguishes the background drift difference between the edge group and the center group. The whole process smoothly transitions from discrete data to continuous distribution, and then to drift quantization and region division, forming a complete technical logic chain. These processing results can effectively identify the drift characteristics caused by thermal gradient and light path refraction on the reagent plate, providing key information support for spatial correction of real-time amplification curves, thereby improving the accuracy and reliability of chicken feed utilization rate trait molecular detection.

[0051] The pre-drift unit takes the initial position-light temperature grid as the reference, performs convolution interpolation operation, outputs the hole position drift prediction surface and drift mask, and distinguishes the background difference between the edge group and the center group. However, during the detection process of real-time amplification curves in the closed channel, the edge hole background fluorescence is lifted due to the influence of thermal radiation and light path refraction difference, which may cause the interpretation program to misjudge the drift as a real amplification signal. To solve this problem, the curve correction unit adds the drift mask to the real-time amplification curve by period weighting to eliminate the background drift component and generate a set of drift corrected curves, ensuring signal accuracy.

[0052] The task of the curve correction unit is to apply the drift mask generated by the pre-drift unit to the real-time amplification curve to eliminate the background drift component and generate a set of drift corrected curves. The following is a detailed technical logic description: Acquisition of real-time amplification curve: The real-time amplification curve is the core data reflecting the dynamic process of nucleic acid amplification reaction, and its acquisition process needs to ensure the accuracy and integrity of the data. The specific steps for acquiring the real-time amplification curve are as follows: during the heating cycle of nucleic acid amplification reaction, when each heating stage ends, the fluorescence intensity of each hole is immediately detected and recorded by the optical module, generating fluorescence values corresponding to that time point. These fluorescence values are arranged in time sequence to form a sequence that changes with time, which is the real-time amplification curve. The fluorescence intensity data of each hole is collected and stored independently to ensure that the data does not interfere with each other, while retaining complete records for subsequent processing. In this way, the real-time amplification curve can accurately reflect the dynamic changes of nucleic acid amplification, providing reliable raw data support for subsequent signal correction and analysis.

[0053] Periodic weighting of drift mask: Considering that the thermal radiation and optical path refraction difference in the closed channel will gradually accumulate and interfere with the fluorescence signal during the heating cycle, the period weighting design of the drift mask aims to adapt to this cumulative effect by dynamically adjusting the correction strength. The specific processing process is as follows: first, determine the period length of the heating cycle, that is, the fixed time interval of each heating cycle; then, for any time point, calculate the number of heating cycles it is in according to the corresponding time, denoted as the current cycle number; then, use the preset drift mask value and the current cycle number to calculate to determine the drift correction factor. The calculation method of the drift correction factor is as follows: first, add 1 to the current cycle number to get a denominator value, then divide the drift mask value by the denominator value, and finally subtract the quotient from 1 to get the drift correction factor. The processing of adding 1 to the denominator is to avoid the denominator being zero at the first cycle, to ensure the stability of the calculation process. The value of the drift correction factor is always between 0 and 1, and it gradually decreases with the increase of the cycle number, which reflects the characteristic of the drift effect accumulating over time. This period weighting method enables the correction strength to dynamically adapt to the trend of drift change, thereby improving the accuracy of fluorescence signal correction.

[0054] Drift correction of real-time amplification curve: Since the fluorescence intensity in the real-time amplification curve may deviate from the true amplification signal due to the influence of drift, the drift correction process adjusts the original data point by point to restore the authenticity of the signal. The specific correction method is as follows: for each time point in the real-time amplification curve, multiply the original fluorescence intensity value at that time point by the corresponding drift correction factor to obtain the corrected fluorescence intensity value. Since the drift correction factor is valued between 0 and 1, its role is to proportionally reduce the interference part of the original fluorescence intensity caused by drift, while retaining the true signal component related to nucleic acid amplification. By correcting the fluorescence intensity at each time point in this way, the real-time amplification curve can effectively eliminate the influence of drift throughout the entire time sequence, thereby making the curve more truly reflect the dynamic process of nucleic acid amplification and improving the reliability and analysis value of the data.

[0055] Generation of drift correction curve set: In order to facilitate subsequent unified management and feature analysis of multiple well data, the generation process of the drift correction curve set integrates the correction results of all wells into a complete data set. The specific generation steps are as follows: collect the fluorescence intensity time series of each well after drift correction to form a set containing data of all wells, named drift correction curve set. The corrected curve of each well is stored independently in the set, maintaining the independence and traceability of the data. This integration method enables the drift correction curve set to completely retain the corrected data of all wells, facilitating subsequent extraction of key features or further evaluation and analysis, thereby providing high-quality input data support for chicken feed utilization trait molecular detection.

[0056] Through the above steps, from the acquisition of real-time amplification curves, to the period weighting processing of drift masks, to the point-by-point correction of real-time amplification curves, and finally to the generation of drift correction curve sets, a complete technical processing chain is formed. This process effectively eliminates the background drift interference caused by the differences in temperature radiation and light path refraction in the closed channel, allowing the fluorescence signal to more truly reflect the situation of nucleic acid amplification reaction. The processed drift correction curve set provides accurate and reliable data support for chicken feed utilization trait molecular detection, improving the accuracy and practicality of the detection results.

[0057] The light unit to the curve correction unit generates a drift prediction surface and a drift mask by collecting the initial position-light temperature grid and performing convolution interpolation operations, and superimposes them on the real-time amplification curve, successfully eliminating the background drift component and generating a drift correction curve set. This series of operations effectively corrects the edge hole background fluorescence uplift problem caused by the thermal gradient and light path refraction difference in the closed channel, ensuring the accuracy of the fluorescence signal. However, the corrected curve set still needs further analysis to quantify the reliability of the amplification reaction, extract key features and generate allele maps, and identify abnormal samples to avoid misjudgment of signal interference affecting the accuracy of subsequent feed conversion estimates. The evaluation unit is the key link in this context, focusing on multi-dimensional feature extraction and machine learning evaluation of the drift correction curve set.

[0058] The task of the evaluation unit is to extract multi-dimensional key features from the drift correction curve set, use the pre-trained machine learning model based on historical data for evaluation, and decide whether to generate an allele map or mark abnormal samples based on the evaluation results. The following is a detailed technical logic description, divided into four sub-steps.

[0059] Key feature extraction: First, we need to extract quantitative indicators from the corrected curve set that can reflect the reliability of the amplification reaction. This is because the reliability of the amplification reaction directly affects the quality of the subsequent genetic data, and quantitative indicators can provide objective evidence for reliability evaluation. Key feature extraction focuses on two indicators: inflection point clarity score and signal persistence ratio, which analyze the curve from the geometric characteristics and energy persistence, respectively, to fully characterize the amplification reaction.

[0060] For the extraction of inflection point clarity score, the process starts with the second derivative calculation on the corrected curve of each well position. The second derivative is used to identify significant points of curve trend change by finding the positions where the second derivative is zero, determining a set of candidate inflection points. These candidate inflection points reflect the turning features of the curve on the time axis. Then, the point with the maximum curvature is selected from the candidate inflection points as the main inflection point. The calculation method of curvature is: take the absolute value of the second derivative, divide by the result of the square of the first derivative and the value of 1 plus. This calculation method can quantify the bending degree of the curve at a certain point, and the greater the bending degree, the more obvious the dynamic change of the amplification reaction at this point. Finally, the curvature value of the main inflection point is defined as the inflection point clarity score, which is used to represent the geometric characteristics of the amplification reaction.

[0061] For the extraction of signal persistence ratio, the process takes the main inflection point as the dividing point to analyze the performance of the curve before and after the main inflection point. The specific steps are: respectively, the curve in the time period before the main inflection point and the time period after the main inflection point is numerically integrated to calculate the area before the inflection point and the area after the inflection point. The signal persistence ratio is defined as the result of the area after the inflection point divided by the area before the inflection point. To avoid the case where the denominator is zero, a very small positive number is added to the denominator. This ratio reflects the persistence of the signal after the main inflection point of the amplification reaction, and embodies the energy persistence of the reaction. The signal persistence ratio extracted in this way can directly quantify the stability of the amplification signal.

[0062] By synchronously extracting the inflection point clarity score and the signal persistence ratio, the spectrum evaluation unit realizes multi-dimensional characterization of the reliability of the amplification reaction. This multi-dimensional quantification method avoids the one-sidedness that may be caused by a single feature, while reducing the need for repeated data reading, thereby improving the processing efficiency and the accuracy of evaluation, and providing a reliable input for subsequent analysis.

[0063] Machine learning model evaluation: After completing the key feature extraction, the machine learning model evaluation stage is entered, and a pre-trained machine learning model is used to quantitatively analyze the reliability of the amplification reaction of each well position. Due to the complexity of the amplification reaction, artificial rules are difficult to cover all abnormal situations comprehensively, while machine learning models based on historical data can accurately distinguish between normal and abnormal reactions by learning feature patterns. The evaluation process takes the inflection point clarity score and the signal persistence ratio as input, and outputs the amplification completeness index, providing a basis for subsequent decision-making.

[0064] The training of the machine learning model is based on a large amount of historical data containing feature pairs of known normal and abnormal amplification reactions and their corresponding reliability labels. By analyzing these historical data, the model learns the feature rules that distinguish between normal and abnormal amplification reactions. During evaluation, the inflection point clarity score and signal persistence ratio of the current well are combined into a feature pair and input into the pre-trained model. The model calculates and outputs the amplification completeness index according to the mapping relationship established during training. The value range of this index is set to 0 to 1, and the higher the value, the stronger the integrity and reliability of the amplification reaction. Such quantification can effectively identify problems such as background drift or abnormal amplification.

[0065] In specific implementation, support vector machine is selected as the classifier. The principle of support vector machine is to find an optimal hyperplane in the feature space to distinguish between normal and abnormal amplification reactions. During training, the model learns the mapping relationship between features and amplification reliability by optimizing the classification interval between support vectors. During evaluation, for each well, the feature pair is input into the trained support vector machine model, and the model outputs the amplification completeness index as the reliability score of the amplification reaction. The application of support vector machine makes the evaluation process have high classification accuracy when dealing with high-dimensional data.

[0066] Through machine learning model evaluation, the evaluation unit realizes accurate quantification of amplification reaction reliability. This data-driven evaluation method improves the sensitivity and consistency of anomaly detection, ensuring the reliability of chicken feed utilization trait molecular detection data and providing high-quality support for subsequent genetic analysis.

[0067] Decision and generation of allele map: After the machine learning model evaluation is completed, the evaluation unit enters the decision and generation of allele map phase, and decides whether to generate an allele map or mark abnormal samples according to the comparison of the amplification completeness index and the preset threshold. Since genetic evaluation requires high-quality amplification data, setting the threshold can effectively filter out reliable reaction results and timely identify abnormal situations to ensure the overall quality of the data.

[0068] For example, set the preset threshold to 0.8. For each well, if the amplification completeness index is greater than 0.8, the amplification reaction is considered reliable, and the inflection point sequence is extracted from the corrected curve to construct the allele map. The extraction method of the inflection point sequence is: based on the points where the second derivative is zero, record the key change points of the curve on the time axis, which reflect the characteristic turning points of the amplification reaction. If the amplification completeness index is less than or equal to 0.8, the amplification reaction is considered abnormal, and the corresponding well is marked as an abnormal sample. Such threshold decision method ensures that only reliable amplification reactions enter subsequent analysis.

[0069] Allele mapping is a comprehensive map that identifies and records allelic variations of an individual by analyzing its DNA sequence. The construction process usually includes the following steps: first, extract DNA samples from individuals; then, use high-throughput sequencing technology (such as Illumina sequencing) to sequence DNA and generate a large number of short read sequences (reads); then, align these reads to the reference genome to determine their location in the genome; then, identify different sites from the reference genome through alignment results, i.e. allelic variations such as single nucleotide polymorphisms (SNPs) or insertions / deletions (Indels); then, annotate these variations and analyze their functional impact, such as whether they are located in coding regions, introns or exons, etc.; finally, integrate all detected variations to form the individual's allele map. This process relies on high-precision sequencing technology and bioinformatics tools to ensure the accuracy and reliability of the results.

[0070] For the wells marked as abnormal samples, the processing process records their numbers and associates them with sample barcodes, generating a review queue list. This list is used for subsequent manual review or re-detection to further confirm the reasons for the abnormalities. In this way, abnormal samples are effectively isolated, avoiding interference with genetic evaluation.

[0071] Through decision-making and generating allele maps, the evaluation unit realizes precise control of the reliability of amplification reactions. The generated high-quality allele maps provide accurate data support for genetic analysis of chicken feed utilization traits, while the marking and review mechanism of abnormal samples enhances the robustness of the detection process, ensuring the overall data quality.

[0072] The evaluation unit extracts the inflection point clarity score and signal persistence ratio from the corrected curve set and uses a machine learning model pre-trained based on historical data to output an amplification completeness index, completing the precise evaluation of the reliability of amplification reactions for each well. When the amplification completeness index exceeds the preset threshold, an allele map is generated based on the inflection point sequence, providing high-quality data for genetic analysis of chicken feed utilization traits; when the amplification completeness index does not reach the threshold, the well is marked as an abnormal sample and included in the review queue, effectively isolating potential misjudgment signals. This process ensures the credibility of chicken feed utilization trait molecular detection data through multi-dimensional feature extraction, data-driven evaluation and threshold decision-making, avoiding the impact of background drift and abnormal amplification on genetic evaluation, and providing reliable technical support for precise breeding.

[0073] The light unit to the evaluation unit through the collection of initial light temperature grid, the implementation of drift correction, extraction of key features and application of machine learning model, generated high-quality allele map and abnormal sample marking. These steps solve the problem of background drift caused by thermal gradient and light refraction in the closed fluorescent typing pipeline, ensuring the reliability of the detection data. However, in order to realize the accurate association of detection results and production performance, it is necessary to accurately map the allele map generated by the evaluation unit with the feeding record and weight gain log, and properly handle the abnormal samples. The task of the abnormal storage unit is to write the allele map and feeding mapping index into the data management platform after one-to-one correspondence, and list the abnormal sample number in the review queue, to ensure the integrity and reliability of the data link.

[0074] The abnormal storage unit aims to accurately match the allele map generated by the evaluation unit with the feeding mapping index, and review and manage abnormal samples. The following is the detailed technical logic, which is divided into three sub-steps.

[0075] Association of allele map and sample barcode: First, the accurate correspondence between the allele map and the sample barcode needs to be established. As the unique identifier of each hole position sample to be tested, the association of sample barcode with allele map can ensure the traceability and uniqueness of the detection results, avoiding data confusion. The processing process is: for each allele map, according to its corresponding hole coordinates, search for the sample barcode matching the hole coordinates in the sample barcode set, and then generate a corresponding record containing the allele map and sample barcode. The whole process takes hole coordinates as a connecting bridge, and completes the matching through one-by-one comparison, ensuring that each allele map can accurately point to a specific sample barcode. This association method provides a clear path for subsequent data integration, while ensuring the integrity of the data link, making it easy to track the source of each detection result.

[0076] Matching of feeding mapping index: After the completion of the corresponding records of allelic profiles and sample barcodes, it is necessary to interface these records with the mapping index of feeding records and weight gain logs to achieve the comprehensive integration of detection data and production performance data. The feeding mapping index stores sample barcodes and their corresponding feeding data and weight gain data, which are the key basis for evaluating chicken feed utilization traits. The processing process is as follows: for each corresponding record containing allelic profiles and sample barcodes, find the entry consistent with the sample barcode in the feeding mapping index, extract the corresponding feeding data and weight gain data, and then generate a complete data combination containing allelic profiles, feeding data and weight gain data. The matching process takes sample barcodes as the core link, and realizes the precise docking of data through item-by-item retrieval. This way ensures the consistency between detection results and production performance data, forming a comprehensive data set, providing high-quality basis for subsequent genetic evaluation and breeding decisions.

[0077] Review queue management of abnormal samples: In order to ensure the reliability and accuracy of detection data, special review queue management is carried out for abnormal samples to identify and handle false positive signals that may affect the results. Abnormal samples refer to those wells whose amplification completeness index does not meet the preset standard, and their detection results may be biased, which need to be further checked or re-detected. The processing process is as follows: for each well identified as abnormal, retrieve the corresponding sample barcode in the sample barcode set according to its well coordinates, generate review records containing sample barcodes and well coordinates, and then integrate these records into the review queue for subsequent processing. The whole management process ensures that abnormal data can be isolated and reviewed in a timely manner by recording detailed information of abnormal samples. This mechanism effectively reduces the interference of abnormal signals on the overall data quality, improves the reliability and stability of the detection process, and provides a guarantee for the accuracy of the final results.

[0078] The abnormal sample unit realizes the precise integration of detection data and production performance data by establishing the correspondence between allelic profiles and sample barcodes, interfacing the feeding mapping index, and managing the review queue of abnormal samples, while effectively isolating abnormal samples. In the application of chicken feed utilization trait molecular detection, each allelic profile can form a traceable complete record with the corresponding feeding data and weight gain data, providing solid data support for genetic evaluation. The review mechanism of abnormal samples reduces the influence of false positives on feed conversion estimates by isolation and verification, ensuring the stability and accuracy of the data link, and providing reliable high-quality data for the breeding goal of feed utilization traits.

[0079] The light unit to the storage unit collects the initial light temperature grid, performs drift correction, extracts key features, applies machine learning models, and integrates detection data and production performance data, completing the whole process from raw signal collection to allele map generation, and effectively identifying abnormal samples. This process ensures accurate mapping of test results and feeding data, providing a reliable foundation for genetic evaluation of feed utilization traits. However, in long-term operation, the thermal gradient and light path refraction in the closed channel may cause hole position drift to gradually accumulate, and if not dynamically adjusted, hidden bias will affect detection accuracy. Therefore, the more unit introduces a recursive thermal light field calibration mechanism, using the current batch's drift prediction surface and correction parameters to optimize the initialization data of the next batch, to achieve continuous improvement of the thermal light field model and long-term stability of detection accuracy.

[0080] The more unit aims to use the current batch's hole position drift prediction surface and its corresponding correction parameters as the initialization data for the next batch, generating a recursive thermal light field calibration template to optimize the thermal light field model and reduce the impact of drift accumulation in long-term operation. The following is the detailed technical logic, divided into three sub-steps, each based on the output of the previous step to ensure the continuity and consistency of data flow.

[0081] Recording of hole position drift prediction surface: First, the hole position drift prediction surface of the current batch needs to be recorded in its entirety to ensure that subsequent batches can use these data for drift correction. The hole position drift prediction surface contains the fluorescence drift and temperature drift of each hole, which comes from the convolution interpolation operation based on the initial position-light temperature grid in the pre-drift unit. The specific process of recording is as follows: take the number of the current batch as the index, store the hole position drift prediction surface in the database of the data management platform, and ensure that the fluorescence drift and temperature drift of each hole are matched with the corresponding hole coordinates. The database can quickly retrieve the drift feature data of the current batch through this structured storage method, and directly call these information when performing drift correction in the next batch. This recording method preserves the link between batches by retaining the drift data link, ensuring that the thermal light field model can continuously use historical information for optimization in long-term operation, while maintaining data traceability.

[0082] Calculation and recording of correction parameters: After the completion of the recording of the well-site drift prediction surface, it is necessary to further calculate and record the average drift correction factor of each well site to quantify the correction requirement of each well site in the detection process. The average drift correction factor is derived from the parameters used to correct the real-time amplification curve in the correction unit, reflecting the average influence of drift in the detection process. The calculation process is as follows: for each well site, the drift correction factors at all time points in the current batch are accumulated, and then divided by the total number of time points to obtain the average drift correction factor of the well site. After the calculation is completed, the average drift correction factor of each well site is stored in the database together with the fluorescence drift and temperature drift in the well-site drift prediction surface, ensuring consistency in the well-site coordinates. In this way, the database not only saves the original drift data, but also provides a quantized correction reference value, enabling the initialization correction of the next batch to be directly adjusted based on these data, thereby improving the optimization effect of the thermal-optical field model.

[0083] Generation of recursive thermal-optical field calibration template: After the completion of the recording of the well-site drift prediction surface and the average drift correction factor, the generation of the recursive thermal-optical field calibration template is entered to optimize the initialization parameters of the next batch through batch data fusion. The recursive thermal-optical field calibration template aims to integrate the drift information of the current batch and the previous batch, and dynamically adapt to changes in long-term operation. The generation process is divided into two cases: if the current batch is the first batch, the well-site drift prediction surface and the average drift correction factor of the current batch are directly used as the initialization table of the next batch; if there is an initialization table of the previous batch, the initialization data of the next batch is updated through weighted average. The specific calculation method is as follows: the well-site drift prediction surface and the average drift correction factor of the previous batch are added to the corresponding data of the current batch, and then summed after assigning weights according to the forgetting factor. The forgetting factor is a value between 0 and 1, used to adjust the relative importance of historical data and current data, and the smaller the value, the greater the influence of current data. After the calculation is completed, the generated initialization table is stored in the database with the batch number as the index, facilitating direct loading of the next batch. Through this dynamic updating mechanism, the thermal-optical field calibration template can gradually adapt to the changes in thermal gradient and optical path refraction in the closed channel, suppress the accumulation of hidden bias, and thus improve the long-term stability and accuracy of the detection results.

[0084] The more modular unit realizes the batch optimization of the thermal optical field model by recording the current batch hole position drift prediction surface, calculating and recording the average drift correction factor, and generating a recursive thermal optical field calibration template. In the application of chicken feed utilization trait molecular detection, this technical logic dynamically adjusts the initial parameters using the actual drift data of each batch, enabling the thermal optical field model to adapt to long-term changes in the detection environment. By suppressing the accumulation of hidden bias, this mechanism ensures the stability and accuracy of the detection results, providing reliable data support for subsequent genetic evaluation and feed utilization improvement.

[0085] The above formulas are dimensionless numerical calculations, and the formulas are obtained by software simulation of a large amount of data to obtain a formula of the latest real situation, and the preset parameters in the formula are set by a person skilled in the art according to the actual situation.

[0086] It should be noted that the system of the present application can be deployed on the device itself to realize embedded application, or can be run on PC or other terminal with user interface, so as to meet various hardware environments and use requirements.

[0087] The above only describes some exemplary embodiments of the present application by way of illustration, and it is needless to say that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present application. Therefore, the above figures and descriptions are illustrative in nature and should not be understood as limiting the scope of protection of the claims of the present application.

[0088] It should be noted that in this paper, if there are relationship terms such as first and second, they are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the term "includes" "contains" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without more limitation, the element defined by the sentence "includes a" does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0089] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, and any person skilled in the art can easily think of changes or replacements within the technical range disclosed in the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A data management system for a chicken feed utilization trait molecular detection kit, characterized in that, The method comprises the steps of: Lighting unit: At the end of sample addition, the initial fluorescence intensity and temperature of all wells on the reagent plate are synchronously collected, and the coordinates are recorded to generate an initial position-light temperature grid; Pre-drift unit: Based on the initial position-light temperature grid, a convolution interpolation operation is performed to output a well drift prediction surface and a corresponding drift mask, and to distinguish the edge group and the center group background; Calibration unit: The drift mask is superimposed on the real-time amplification curve by period weighting, and the drift correction curve set is obtained after removing the background drift component; Spectrum evaluation unit: Extracting multi-dimensional key features from the drift correction curve set, inputting the historical data pre-trained machine learning model to evaluate the amplification completeness, and generating an allele spectrum or marking abnormal samples according to the evaluation result; Store unit: The allele spectrum is correspondingly written into the data management platform after being mapped with the feeding index, and the abnormal sample number is included in the review queue; More module unit: The well drift prediction surface and the corresponding correction parameters are written into the next batch initialization table to be used as a recursive thermal light field calibration template.

2. The chicken feed utilization trait molecular detection kit data management system according to claim 1, wherein, The lighting unit runs as follows: After the reaction liquid sample is added, the optical module and the temperature sensor are started immediately to synchronously scan all wells on the reagent plate, and the initial fluorescence intensity and the initial temperature of each well are recorded; Each well on the reagent plate is assigned a unique two-dimensional coordinate as its spatial position identifier; the initial fluorescence intensity and the initial temperature of each well are combined into a data pair, and a two-dimensional matrix is formed according to the two-dimensional coordinates to generate an initial position-light temperature grid; the initial position-light temperature grid is stored in the database to ensure that the data pair of each well is bound to its two-dimensional coordinates.

3. The chicken feed utilization trait molecular detection kit matching data management system according to claim 2, characterized in that, The pre-drift unit runs as follows: Based on the initial position-light temperature grid, a Gaussian convolution kernel is used to perform interpolation operation on the fluorescence intensity and temperature data to generate continuous fluorescence intensity prediction surface and temperature prediction surface; the predicted value of each well is extracted from the fluorescence intensity prediction surface and the temperature prediction surface to calculate the fluorescence drift and the temperature drift to generate the well drift prediction surface; based on the well drift prediction surface, the drift mask value of each well is calculated to identify the drift severity; the wells are divided into an edge group and a center group, and the average drift mask value of the edge group and the center group is calculated to quantify the background drift difference.

4. The chicken feed utilization trait molecular detection kit complete data management system according to claim 3, characterized in that, The calibration unit runs as follows: The fluorescence intensity of each well is captured immediately after each heating cycle by the optical module to obtain real-time amplification curves and generate time series data; the drift correction factor is calculated according to the period length of the heating cycle and the current cycle number, and the correction intensity is dynamically adjusted by period weighting of the drift mask.

5. The chicken feed utilization trait molecular detection kit data management system according to claim 4, wherein, The calibration unit also runs as follows: The drift correction factor is applied to the original fluorescence intensity at each time point to perform drift correction of the real-time amplification curve to generate corrected fluorescence intensity time series; the corrected curves of all wells are integrated into a drift correction curve set to form a complete data set.

6. The chicken feed utilization trait molecular detection kit data management system according to claim 5, wherein, The spectrum evaluation unit runs as follows: Extract the corrected fluorescence intensity time series of each well position from the drift correction curve set; calculate the second derivative of the corrected fluorescence intensity time series of each well position to identify the candidate inflection point set, and select the inflection point with the maximum curvature as the main inflection point, and record its curvature value as the inflection point clarity score; with the main inflection point as the boundary, calculate the area before the inflection point and the area after the inflection point, and then calculate the signal persistence ratio; input the inflection point clarity score and signal persistence ratio of each well position as features into the support vector machine model pre-trained based on historical data, and output the amplification integrity index.

7. The chicken feed utilization trait molecular detection kit data management system according to claim 6, wherein, The evaluation unit operation further includes the following: According to the comparison of the amplification integrity index and the preset threshold value, it is determined whether to extract the inflection point sequence from the corrected curve to generate the allele graph, or to mark the well position as an abnormal sample and record the number.

8. The chicken feed utilization trait molecular detection kit data management system according to claim 7, wherein, The abnormality storage unit operation includes the following: For each allele graph, search for the matching sample barcode in the sample barcode set according to its corresponding well position coordinates, generate a corresponding record containing the allele graph and the sample barcode; for each corresponding record containing the allele graph and the sample barcode, find the entry consistent with the sample barcode in the feeding mapping index, extract the corresponding feeding data and weight gain data, and generate a complete data combination containing the allele graph, the feeding data and the weight gain data; for each well position identified as abnormal, search for the corresponding sample barcode in the sample barcode set according to its well position coordinates, generate a review record containing the sample barcode and the well position coordinates, and integrate the review record into the review queue.

9. The chicken feed utilization trait molecular detection kit data management system according to claim 8, wherein, The more module unit operation includes the following: Store the well position drift prediction surface of the current batch to the database, and record the fluorescence drift and temperature drift of each well position with the batch number as the index; for each well position, calculate the average value of its drift correction factor at all time points in the current batch to obtain the average drift correction factor, and store it together with the well position drift prediction surface.

10. The chicken feed utilization trait molecular detection kit data management system according to claim 9, wherein, The more module unit operation further includes the following: According to whether the current batch is the first batch, if the current batch is the first batch, directly use the well position drift prediction surface and the average drift correction factor as the initialization table of the next batch, if there is an initialization table of the last batch, fuse the well position drift prediction surface, the average drift correction factor of the last batch and the data of the current batch by weighted average to generate the recursive thermal light field calibration template of the next batch.

Images