A method for identifying drought-resistant genes in potatoes using hyperspectral

Abstract

The application provides a method for identifying potato drought-resistant genes using hyperspectral, and relates to the technical field of crop drought-resistant gene identification, and comprises the following steps: setting a drought group and a control group, synchronously acquiring plant canopy hyperspectral images and transcriptome data and establishing a correlation data set; extracting spectral features, screening drought response candidate genes; calculating the correlation coefficient of each candidate gene and the spectral features based on Spearman correlation analysis, and determining the primary gene set according to whether the average value of a plurality of positions after the absolute value is sorted exceeds a threshold value; constructing a random forest regression model with the spectral features as input and the expression amount of the primary gene as output, predicting the expression amount through a verification sample, and screening target drought-resistant genes according to the coefficient of determination and the prediction error; constructing a target gene overexpression plant, measuring the relative change rate of physiological indexes and the change multiple of gene expression amount under drought, calculating a comprehensive function verification index, and determining the gene as a drought-resistant gene if the index exceeds a threshold value.

Description

Technical Field

[0001] This invention relates to the field of crop drought resistance gene identification technology, specifically a method for identifying potato drought resistance genes using hyperspectral imaging. Background Technology

[0002] Potatoes are a vital global food and economic crop, and their stable production is crucial for ensuring food security. However, drought stress is one of the major abiotic stresses affecting potato yield and quality. Traditional methods for identifying drought-resistant genes in potatoes mainly rely on destructive measurements of physiological and biochemical indicators and functional verification based on gene cloning and genetic transformation. While these methods are mature, they have limitations such as low throughput, long cycles, significant plant damage, and difficulty in achieving large-scale, non-destructive screening, thus hindering the rapid discovery and breeding application of drought-resistant genes. Therefore, developing a method that can non-destructively, rapidly, and with high throughput correlate macroscopic drought phenotypes with microscopic gene expression in plants and accurately identify drought-resistant genes is of great value for potato drought-resistant breeding research.

[0003] In the prior art, a method for identifying drought-resistant genes in maize using hyperspectral imaging (CN111781150A) is disclosed. This method combines high-throughput hyperspectral detection technology with genome-wide association analysis (GWAS). It mines candidate drought-resistant genes in maize through non-destructive acquisition of hyperspectral phenotypic traits, dimensionality reduction screening, heritability analysis, and GWAS analysis. While this method achieves non-destructive high-throughput acquisition of phenotypic traits, its technical approach mainly relies on population genetics and statistical genetics analysis. GWAS is used to locate genomic regions significantly associated with the phenotype, thereby screening candidate genes. However, GWAS methods typically struggle to directly and precisely establish a quantitative association model between real-time, dynamic macroscopic spectral phenotypic characteristics of plants under specific environmental stresses and the expression levels of specific genes within them. The candidate genes screened are based on statistical associations between population genetic variation and phenotypic variation, rather than a direct causal model based on gene expression and spectral characteristics. Furthermore, this method lacks the crucial step of constructing predictive models to directly and quantitatively predict the expression levels of genes that are difficult to measure using easily measurable spectral features. This is essential for achieving rapid and non-destructive genotype inference and functional pre-screening. Therefore, there is an urgent need for a new method that can more directly link phenotype and genotype under stress and construct predictive models to make up for the shortcomings of existing technologies.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a method for identifying drought-resistant genes in potatoes using hyperspectral imaging, in order to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A method for identifying drought-resistant genes in potatoes using hyperspectral imaging, comprising the following steps:

[0008] Set up drought stress group and control group for potato plants, and simultaneously acquire hyperspectral image data of the canopy and transcriptome expression profile data of each group of plants to construct an associated dataset;

[0009] A spectral feature set including differential vegetation index, photochemical reflectance index, water stress index and red edge position parameter was extracted from hyperspectral image data, and a candidate gene expression level dataset in response to drought stress was screened from transcriptome expression profile data.

[0010] Based on Spearman rank correlation analysis, the correlation coefficient between each feature in the spectral feature set and the expression level of each candidate gene in the candidate gene expression dataset is calculated. For each candidate gene, the absolute values ​​of its correlation coefficients with all spectral features are sorted in descending order. Candidate genes are then selected and included in the initial gene set based on the sorting.

[0011] Each drought-resistant plant was randomly divided into a training set and a validation set. The spectral feature set in the training set was used as the feature input, and the expression level of each gene in the initial gene set was used as the feature output to train the random forest regression model. The spectral features of the validation set were input into the trained model to obtain the predicted expression level of each gene. The determination coefficient between the predicted expression level and the actual expression level of each gene was calculated. Based on the determination coefficient, the target drought-resistant genes were selected.

[0012] Using the target drought-resistant gene as the object, transgenic plants overexpressing it were constructed. The relative change rate of key physiological indicators of transgenic plants relative to wild-type plants under drought stress was measured. Combined with the fold change of the expression level of the target drought-resistant gene in the transgenic plants, it was determined whether the gene is a potato drought-resistant gene with significant drought resistance function.

[0013] Furthermore, the process of establishing the associated dataset is as follows:

[0014] Potato plants from the same batch, at the same growth stage, and with consistent physiological states were selected. One plant was randomly chosen as the normal water supply control group and maintained under normal water supply. The remaining plants were treated as the drought stress group. At the time point when the experimental treatment reached the predetermined stress level, hyperspectral imagers were used to simultaneously acquire the reflectance spectra of the canopy of each group of living plants under fixed illumination conditions. Within one hour of the spectral acquisition, fresh leaf samples from the same physiological part of each group of plants were taken, and RNA was immediately extracted and sequenced to obtain their transcriptome expression profile data. The hyperspectral image data of each plant was uniquely associated with its own transcriptome expression profile data to form a one-to-one associated dataset.

[0015] Furthermore, the hyperspectral image data is preprocessed by radiometric calibration, atmospheric correction and geometric correction in sequence. For the preprocessed hyperspectral image data, the average reflectance spectral curve of the effective imaging area of ​​each plant canopy is extracted, and the difference vegetation index, photochemical reflectance index, water stress index and red edge position parameter are calculated based on the spectral curve to form a spectral feature set.

[0016] The formula for calculating the difference vegetation index is as follows:

[0017] ;

[0018] In the formula, The difference is the vegetation index; The average reflectance in the near-infrared band; The average reflectance of the red band;

[0019] The formula for calculating the photochemical reflectance index is as follows:

[0020] ;

[0021] In the formula, Photochemical reflectance index; , They are respectively At wavelength and Reflectivity at wavelength;

[0022] The formula for calculating the water stress index is as follows:

[0023] ;

[0024] In the formula, The water stress index; The average reflectance of the shortwave infrared band 1;

[0025] The red-edge position parameter is calculated by measuring the spectral curve. The wavelength corresponding to the maximum value of the first derivative within the red-edge region is determined;

[0026] Simultaneously, from the transcriptome expression profile data, the expression levels of each gene in each drought-stressed plant in the drought stress group and the normal water supply control group are compared in pairs. The fold change in the expression level of each gene between the two groups is calculated. Genes with an absolute fold change in expression level greater than a preset fold threshold are selected as candidate genes. The candidate gene expression level data are then compiled to form a candidate gene expression level dataset.

[0027] Furthermore, based on Spearman's rank correlation analysis, the correlation coefficient between each feature in the spectral feature set and the expression level of each candidate gene in the candidate gene expression dataset is calculated. The specific logic is as follows:

[0028] For the first in the candidate gene expression level dataset The candidate gene, and the first candidate gene in the spectral feature set. Using spectral features, the expression level sequence of this gene on all drought-stricken plants in the associated dataset was obtained, respectively. and the sequence of characteristic values ​​of this spectral feature on the corresponding plant. ,in This is an index of candidate genes in the candidate gene expression level dataset. This serves as an index for spectral features, including the difference vegetation index, photochemical reflectance index, water stress index, and red edge position parameter. This represents the number of drought-stricken plants in the associated dataset. An index for drought-affected plants;

[0029] Based on expression level sequences Construct expression level sequence Based on eigenvalue sequences Constructing eigenvalue level sequences ,in, for In sequence Ascending order ranking in the middle for In sequence The values ​​are ranked in ascending order, and if there are ties, the average ranking value is used.

[0030] Furthermore, the correlation coefficient between each spectral feature in the spectral feature set and the expression level of each candidate gene in the candidate gene expression dataset is calculated using the following formula:

[0031] ;

[0032] ;

[0033] In the formula, Indicates the first The candidate gene and the first Spearman's rank correlation coefficient for each spectral feature; Indicates the first On the first drought-stricken plant The expression levels of each candidate gene are related to the first... The difference in the grade of each spectral feature value;

[0034] For the For each candidate gene, obtain the set of absolute values ​​of its correlation coefficients with all spectral features. Sort the values ​​in this set in descending order, and select the top [values] from the sorted sequence. The absolute value of each correlation coefficient is used to calculate this. arithmetic mean of the values ,Will With preset threshold If a comparison is made, Then the first One candidate gene was included in the initial gene set.

[0035] Furthermore, the expression level of each gene in the initial gene set is used as the regression prediction target, and all spectral feature values ​​in the spectral feature set are used as input features to construct a random forest regression model. First, each drought-stricken plant is randomly divided into a training set and a validation set. The training set is used to train the random forest regression model, and the validation set is used to evaluate the predictive performance of the model.

[0036] In the training set, each spectral feature value in the spectral feature set is standardized to obtain standardized spectral feature values. The expression level of each gene in the initial gene set is used as the corresponding regression output target and input into the random forest regression model for training.

[0037] Using a pre-trained random forest regression model, for each plant individual in the validation set, its corresponding standardized spectral feature value is input to obtain the predicted expression level of each gene in the validation set samples. For each gene, the coefficient of determination between its predicted expression level and its actual expression level in the validation set samples is calculated using the following formula:

[0038] ;

[0039] In the formula, The coefficient of determination; Indicates the initial selection of genes in the gene pool. In the verification set The true expression level of each plant. To verify the index of the concentrated plants; Indicates the initial selection of genes in the gene pool. In the verification set Predicted expression levels for each plant; Indicates the initial selection of genes in the gene pool. The average value of the true expression level in the validation set; To verify the number of plants collected;

[0040] The prediction error for each gene in the initial gene set is calculated using the following formula:

[0041] ;

[0042] in, Indicates the initial selection of genes in the gene pool. The prediction error;

[0043] The calculated coefficient of determination With the preset first threshold Compare and set the prediction error With the preset second threshold Compare, if satisfied and Then the initially selected gene The target drought-resistant gene was identified.

[0044] Furthermore, for each target drought-resistant gene, corresponding overexpressing transgenic potato plants were constructed, and wild-type plants with the same genetic background were used as controls. Both the transgenic plant group and the wild-type plant group were simultaneously subjected to drought stress treatment consistent with the conditions set in step S1. After the treatment reached the predetermined stress level, the key physiological indicators of each plant were measured. The key physiological indicators included relative leaf water content, net photosynthetic rate, malondialdehyde content, and proline content.

[0045] The relative rate of change of each key physiological indicator between the transgenic plant group and the wild-type plant group was calculated. At the same time, the expression level of the target drought resistance gene in each plant was measured, and the fold change of the expression level of the target drought resistance gene in the transgenic plant relative to the wild-type plant was calculated.

[0046] Among them, for each key physiological indicator The formula for calculating its relative rate of change is as follows:

[0047] ;

[0048] In the formula, Indicates the first The relative rate of change of key physiological indicators It serves as an index for key physiological indicators. For the first The arithmetic mean of the measured values ​​of several key physiological indicators in all plants in the transgenic plant group. For the first The arithmetic mean of the values ​​of the key physiological indicators measured in all plants in the wild-type plant group;

[0049] The formula used to calculate the fold change in expression level is as follows:

[0050] ;

[0051] In the formula, The change in expression level is a factor of 5. The expression level of the target drought-resistant gene is the arithmetic mean of the expression levels of all plants in the transgenic plant group. The expression level of the target drought-resistant gene is the arithmetic mean of the expression levels of all plants in the wild-type plant group;

[0052] Based on the relative rate of change and the fold change in expression level mentioned above, the comprehensive functional validation index is determined using the following formula:

[0053] ;

[0054] In the formula, The comprehensive functional verification index; , , , These represent the relative changes in leaf relative water content, net photosynthetic rate, malondialdehyde content, and proline content, respectively. , , , , Assign a preset weight to the corresponding indicator, and satisfy the following conditions: ;

[0055] The calculated comprehensive functional verification index Compared with the preset third threshold If a comparison is made, If the target drought-resistant gene is identified as a potato drought-resistant gene with significant drought-resistant function, then the target drought-resistant gene is determined to be a potato drought-resistant gene.

[0056] Compared with the prior art, the beneficial effects of the present invention are:

[0057] First, this invention achieves a direct correlation between macroscopic, non-destructive spectral phenotypes and microscopic, precise gene expression profiles of plants by simultaneously acquiring hyperspectral image data and transcriptome expression profile data of the same plant under drought stress and establishing a one-to-one corresponding association dataset. This method breaks through the limitations of traditional methods that rely on the measurement of destructive physiological indicators or the indirectness of population statistical association (such as GWAS). It can capture the real-time dynamic response relationship between phenotype and genotype under drought stress more directly and precisely, providing a solid data foundation for accurately identifying key genes directly related to drought response.

[0058] Secondly, this invention creatively introduces and constructs a machine learning prediction model. It uses Spearman correlation analysis for initial screening, and then trains a random forest regression model using the selected spectral feature set as input and candidate gene expression levels as output. The model performance is validated and genes are further screened using the coefficient of determination and prediction error. The core value of this technical approach lies in the successful establishment of a computational model that can quantitatively predict the expression levels of specific genes that are difficult to measure directly and have low throughput, using readily available, high-throughput canopy spectral data. This provides a powerful tool for non-destructive, rapid, and high-throughput screening and prediction of drought-resistant gene status in potato plants, significantly improving the efficiency and targeting of gene mining.

[0059] Furthermore, after identifying the target drought-resistant gene through a machine learning model, this invention further constructs overexpressing transgenic plants, measures key physiological indicators and changes in candidate gene expression levels under drought stress, and calculates a comprehensive functional verification index for final biological verification. This not only ensures the scientific rigor and interpretability of the candidate gene screening process, but also confirms the drought-resistant function of the gene through rigorous transgenic functional experiments, ensuring the reliability of the identification results. Attached Figure Description

[0060] Figure 1 This is a schematic diagram of the overall method flow of the present invention;

[0061] Figure 2 The graph shows the functional relationship between the relative change rate of net photosynthetic rate and the comprehensive functional verification index.

[0062] Figure 3 This is a parallel coordinate graph of the relative change rate of each key physiological indicator and the comprehensive functional verification index. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0064] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0065] Example:

[0066] Please see Figures 1-3 The present invention provides a technical solution:

[0067] A method for identifying drought-resistant genes in potatoes using hyperspectral imaging, comprising the following steps:

[0068] S1, set up drought stress group and control group of potato plants, and simultaneously acquire hyperspectral image data of canopy and transcriptome expression profile data of plants in each group to construct an associated dataset;

[0069] In this embodiment, the process of establishing the associated dataset is as follows:

[0070] Potato plants from the same batch, at the same growth stage, and with consistent physiological states were selected. One plant was randomly chosen as the normal water supply control group and maintained under normal water supply. The remaining plants were treated as the drought stress group. At the time point when the experimental treatment reached the predetermined stress level, hyperspectral imagers were used to simultaneously acquire the reflectance spectra of the canopy of each group of living plants under fixed illumination conditions. Within one hour of the spectral acquisition, fresh leaf samples from the same physiological part of each group of plants were taken, and RNA was immediately extracted and sequenced to obtain their transcriptome expression profile data. The hyperspectral image data of each plant was uniquely associated with its own transcriptome expression profile data to form a one-to-one associated dataset.

[0071] In this embodiment, a gradual soil moisture stress was applied to the drought-stressed group. The relative soil moisture content was gradually reduced to the target stress level by a precisely controlled weighing method, and the target stress level was set at field capacity. The control group with normal water supply maintained the relative soil moisture content at field capacity. Within the specified range, on the day the drought stress reaches the predetermined level, from 9:00 AM to 11:00 AM, under constant indoor light or uniform outdoor natural light conditions, a pushbroom hyperspectral imaging system was used to simultaneously acquire reflectance spectral images of the area directly above the canopy of each living plant. Within one hour of completing the hyperspectral acquisition, approximately the middle portion of the third fully unfolded leaf from the top of each plant was immediately harvested. Fresh samples were rapidly immersed in liquid nitrogen for freezing, and the total amount was completed within the same batch. Extraction, quality control, and transcriptome sequencing were performed. Finally, a unique identifier was used to precisely match the hyperspectral image data file of each plant with its corresponding transcriptome expression profile data file, thus constructing a one-to-one associated dataset.

[0072] S2, extract spectral feature sets including differential vegetation index, photochemical reflectance index, water stress index and red edge position parameters from hyperspectral image data, and screen candidate gene expression level datasets in response to drought stress from transcriptome expression profile data;

[0073] In this embodiment, the hyperspectral image data is preprocessed by radiometric calibration, atmospheric correction and geometric correction in sequence. For the preprocessed hyperspectral image data, the average reflectance spectral curve of the effective imaging area of ​​each plant canopy is extracted, and the difference vegetation index, photochemical reflectance index, water stress index and red edge position parameter are calculated based on the spectral curve to form a spectral feature set.

[0074] The formula for calculating the difference vegetation index is as follows:

[0075] ;

[0076] In the formula, The difference is the vegetation index; for The average reflectance in the near-infrared band; for The average reflectance in the red band; the difference vegetation index is an important spectral indicator reflecting vegetation growth and biomass. A value closer to 1 indicates high vegetation cover, lush foliage, abundant chlorophyll content, and active photosynthesis; conversely, a value closer to 1 indicates lower vegetation cover. Values ​​close to 0 or negative indicate sparse vegetation, wilted leaves, chlorophyll degradation, or impaired photosynthesis; under drought stress, plants typically exhibit the following characteristics: The decrease reflects drought response characteristics such as reduced leaf water content, decreased leaf area, or weakened photosynthetic capacity.

[0077] The formula for calculating the photochemical reflectance index is as follows:

[0078] ;

[0079] In the formula, Photochemical reflectance index; , They are respectively At wavelength and Reflectance at a given wavelength; the photochemical reflectance index is a spectral indicator used to characterize the photosynthetic efficiency and photoprotective capacity of plants. An elevated value usually indicates enhanced photosynthetic activity and improved light energy utilization efficiency; while A decrease in the value may reflect damage to the photosystem, increased photoinhibition, or the activation of photoprotective mechanisms, especially under stress conditions; in arid environments, The decrease is often associated with damage to photosynthetic structures and enhanced non-photochemical quenching, and can serve as a marker of photosynthetic response to early drought stress.

[0080] The formula for calculating the water stress index is as follows:

[0081] ;

[0082] In the formula, The water stress index; for The average reflectance of the shortwave infrared band 1; the water stress index is an indicator that assesses plant water status by comparing the reflectance of the shortwave infrared band to that of the near-infrared band. A higher value indicates lower leaf water content and greater damage to cell structure, reflecting a stronger state of water stress in the plant; conversely, a lower value indicates lower water content. A lower value indicates higher leaf water content and better moisture status; under drought conditions, The increase in γ directly reflects leaf water loss, cell dehydration, and possible membrane system damage, and is a key spectral feature for evaluating drought severity and plant water status.

[0083] The red-edge position parameter is calculated by measuring the spectral curve. The wavelength corresponding to the maximum value of the first derivative within the red-edge region is determined. The red-edge position parameter is used to characterize the chlorophyll content and physiological state of leaves. A shift of the red-edge position towards longer wavelengths usually indicates an increase in chlorophyll content, compact leaf structure, and vigorous growth. A shift towards shorter wavelengths is often accompanied by chlorophyll degradation, leaf senescence, or stress response. Under drought stress, the red-edge position often undergoes a blue shift, reflecting chlorophyll loss, decreased photosynthetic function, and leaf structure damage. It is an important spectral parameter for monitoring physiological decline caused by drought.

[0084] Simultaneously, from the transcriptome expression profile data, the expression levels of each gene in each drought-stressed plant in the drought stress group and the normal water supply control group are compared in pairs. The fold change in the expression level of each gene between the two groups is calculated. Genes with an absolute fold change in expression level greater than a preset fold threshold are selected as candidate genes. The candidate gene expression level data are then compiled to form a candidate gene expression level dataset.

[0085] The preset fold change threshold is set to 2.0. In transcriptomics studies, a fold change in expression greater than 2 is generally considered to have significant biological significance. It can effectively distinguish genes that are significantly upregulated or downregulated due to drought stress, while excluding non-specific expression changes caused by experimental errors or natural fluctuations, thereby ensuring the reliability and specificity of the candidate gene set.

[0086] S3, based on the Spearman rank correlation analysis method, calculates the correlation coefficient between each feature in the spectral feature set and the expression level of each candidate gene in the candidate gene expression dataset. For each candidate gene, sorts the absolute values ​​of its correlation coefficients with all spectral features in descending order, and selects candidate genes and includes them in the initial gene set according to the sorting.

[0087] In this embodiment, based on the Spearman rank correlation analysis method, the correlation coefficient between each feature in the spectral feature set and the expression level of each candidate gene in the candidate gene expression level dataset is calculated. The specific logic is as follows:

[0088] For the first in the candidate gene expression level dataset The candidate gene, and the first candidate gene in the spectral feature set. Using spectral features, the expression level sequence of this gene on all drought-stricken plants in the associated dataset was obtained, respectively. and the sequence of characteristic values ​​of this spectral feature on the corresponding plant. ,in This is an index of candidate genes in the candidate gene expression level dataset. This serves as an index for spectral features, including the difference vegetation index, photochemical reflectance index, water stress index, and red edge position parameter. This represents the number of drought-stricken plants in the associated dataset. An index for drought-affected plants;

[0089] Based on expression level sequences Construct expression level sequence Based on eigenvalue sequences Constructing eigenvalue level sequences ,in, for In sequence Ascending order ranking in the middle for In sequence The values ​​are ranked in ascending order, and if there are ties, the average ranking value is used.

[0090] The correlation coefficient between each spectral feature in the spectral feature set and the expression level of each candidate gene in the candidate gene expression dataset is calculated using the following formula:

[0091] ;

[0092] ;

[0093] In the formula, Indicates the first The candidate gene and the first Spearman's rank correlation coefficient for each spectral feature; Indicates the first On the first drought-stricken plant The expression levels of each candidate gene are related to the first... The difference in the grade of each spectral feature value;

[0094] For this formula, the dependent variable Value between between, Used to measure the strength and direction of the monotonic correlation between gene expression levels and spectral characteristic values, if A value close to 1 indicates a significant positive monotonic relationship between the two, meaning that gene expression increases synchronously with increasing spectral characteristic values; if... If the value is close to -1, it indicates a significant negative monotonic relationship, meaning the two trends are opposite; if... A value close to 0 indicates a lack of monotonic correlation between the two, meaning there is no clear pattern to the changes in gene expression and spectral characteristics; therefore... The magnitude of the value reflects the degree of synergistic change between gene expression and spectral characteristics under drought stress, and is an important basis for screening genes closely related to phenotypic responses.

[0095] Drought stress can cause synchronous changes in plant physiological state and spectral reflectance characteristics. For example, reduced leaf moisture may lead to decreased near-infrared reflectance and increased red light absorption. If the expression response of a certain gene is highly synchronized with this spectral change, its expression ranking in different plants will tend to be consistent with its spectral characteristic ranking. The smaller the spectral response, the greater the difference in ranking. Conversely, if gene expression and spectral response are out of sync, the ranking will differ significantly. Increase; when The smaller the value, the higher the consistency between gene expression and spectral characteristics among plants. The closer the expression is to 1, the stronger the synergy between the two in drought response; conversely, the closer the expression is to 1, the weaker the synergy between the two in drought response. The larger, The closer the value is to 0 or even negative, the more it reflects that the response mechanisms of the two may be separated or have an inverse correlation.

[0096] For the For each candidate gene, obtain the set of absolute values ​​of its correlation coefficients with all spectral features. Sort the values ​​in this set in descending order, and select the top [values] from the sorted sequence. The absolute value of each correlation coefficient is used to calculate this. arithmetic mean of the values ,Will With preset threshold If a comparison is made, Then the first One candidate gene was included in the initial gene set.

[0097] Among them, the preset threshold is determined. The method is as follows: A null distribution is constructed using statistical simulation methods, such as random permutation. This involves randomly shuffling the candidate gene expression sequences while keeping the spectral characteristic value sequences unchanged, and repeatedly calculating the average of the absolute values ​​of the correlation coefficients. The upper quantile of this average at a given significance level is then used as the basis for the analysis. The basis for its value.

[0098] S4. Each drought-stricken plant is randomly divided into a training set and a validation set. The spectral feature set in the training set is used as the feature input, and the expression level of each gene in the initial gene set is used as the feature output to train the random forest regression model. The spectral features of the validation set are input into the trained model to obtain the predicted expression level of each gene. The determination coefficient between the predicted expression level and the actual expression level of each gene is calculated. Based on the determination coefficient, the target drought-resistant genes are selected.

[0099] In this embodiment, the expression level of each gene in the initial gene set is used as the regression prediction target, and all spectral feature values ​​in the spectral feature set are used as input features to construct a random forest regression model. First, each drought-stricken plant is randomly divided into a training set and a validation set. The training set is used to train the random forest regression model, and the validation set is used to evaluate the predictive performance of the model.

[0100] In the training set, each spectral feature value in the spectral feature set is standardized to obtain standardized spectral feature values. The expression level of each gene in the initial gene set is used as the corresponding regression output target and input into the random forest regression model for training.

[0101] Using a pre-trained random forest regression model, for each plant individual in the validation set, its corresponding standardized spectral feature value is input to obtain the predicted expression level of each gene in the validation set samples. For each gene, the coefficient of determination between its predicted expression level and its actual expression level in the validation set samples is calculated using the following formula:

[0102] ;

[0103] In the formula, The coefficient of determination; Indicates the initial selection of genes in the gene pool. In the verification set The true expression level of each plant. To verify the index of the concentrated plants; Indicates the initial selection of genes in the gene pool. In the verification set Predicted expression levels for each plant; Indicates the initial selection of genes in the gene pool. The average value of the true expression level in the validation set; To verify the number of plants collected.

[0104] For this formula, Used to measure the effectiveness of a random forest regression model on initially selected genes. To validate the predictive effectiveness of expression levels in the set. The closer the value is to 1, the better the model's prediction of expression levels matches the actual expression levels, meaning the model effectively captures the relationship between spectral features and gene expression levels, resulting in high prediction accuracy. Conversely, if the value is less than 1, the model's prediction is less accurate. A value close to 0 indicates poor model prediction performance, a weak correlation between predicted and actual values, and that the model failed to effectively utilize spectral features to explain gene expression variations. Under drought stress, a higher value indicates... This means that the expression response of this gene is highly consistent with the changes in spectral phenotype, and it may be involved in drought response mechanisms closely related to spectral characteristics, thus possessing the potential to be a target drought-resistant gene.

[0105] The prediction error for each gene in the initial gene set is calculated using the following formula:

[0106] ;

[0107] in, Indicates the initial selection of genes in the gene pool. The prediction error, expressed as root mean square error, is used to quantify the impact of the random forest regression model on genes. The mean bias of expression level prediction The smaller the value, the smaller the average difference between the model's predicted values ​​and the actual values, the higher the prediction accuracy, and the stronger the model's stability and reliability; conversely, if... A large value indicates a significant deviation between the predicted and actual values, meaning the model's prediction performance is unsatisfactory.

[0108] The calculated coefficient of determination With the preset first threshold Compare and set the prediction error With the preset second threshold Compare, if satisfied and Then the initially selected gene The target drought-resistant gene is identified, wherein the first threshold is... It is determined by combining empirical settings with validation set performance.

[0109] If genes Prediction error Larger, coefficient of determination The low level indicates that the random forest regression model based on spectral features cannot effectively predict the expression level of this gene, reflecting the gene's... There is a lack of stable and reproducible quantitative association between the expression pattern of this gene and the canopy spectral phenotype, meaning that its expression changes under drought stress do not form a significant synergistic response with the physiological state of the plant. Therefore, this gene is weakly associated with key physiological processes under drought stress and does not have the potential to be a target drought-resistant gene. Its function may not be related to the mechanism of directly responding to drought and regulating drought-resistant traits. This means that it is difficult to achieve non-destructive, high-throughput phenotypic-assisted selection of this gene through spectral technology in practical breeding applications, so it is excluded.

[0110] S5. Using the target drought-resistant gene as the object, transgenic plants overexpressing it were constructed. The relative change rate of key physiological indicators of transgenic plants relative to wild-type plants under drought stress was measured. Combined with the fold change of the expression level of the target drought-resistant gene in the transgenic plants, it was determined whether the gene is a potato drought-resistant gene with significant drought resistance function.

[0111] In this embodiment, for each target drought-resistant gene, a corresponding overexpression transgenic potato plant was constructed, and a wild-type plant with the same genetic background was used as a control. Both the transgenic plant group and the wild-type plant group were simultaneously subjected to drought stress treatment consistent with the conditions set in step S1. After the treatment reached the predetermined stress level, the key physiological indicators of each plant were measured. The key physiological indicators included relative leaf water content, net photosynthetic rate, malondialdehyde content and proline content.

[0112] The relative rate of change of each key physiological indicator between the transgenic plant group and the wild-type plant group was calculated. At the same time, the expression level of the target drought resistance gene in each plant was measured, and the fold change of the expression level of the target drought resistance gene in the transgenic plant relative to the wild-type plant was calculated.

[0113] Among them, for each key physiological indicator The formula for calculating its relative rate of change is as follows:

[0114] ;

[0115] In the formula, Indicates the first The relative rate of change of key physiological indicators It serves as an index for key physiological indicators. For the first The arithmetic mean of the measured values ​​of several key physiological indicators in all plants in the transgenic plant group. For the first The arithmetic mean of the values ​​of the key physiological indicators measured in all plants in the wild-type plant group;

[0116] The formula used to calculate the fold change in expression level is as follows:

[0117] ;

[0118] In the formula, The change in expression level is a factor of 5. The expression level of the target drought-resistant gene is the arithmetic mean of the expression levels of all plants in the transgenic plant group. The expression level of the target drought-resistant gene is the arithmetic mean of the expression levels of all plants in the wild-type plant group;

[0119] Based on the relative rate of change and the fold change in expression level mentioned above, the comprehensive functional validation index is determined using the following formula:

[0120]

[0121] In the formula, The comprehensive functional verification index; , , , These represent the relative changes in leaf relative water content, net photosynthetic rate, malondialdehyde content, and proline content, respectively. , , , , The preset weights for the corresponding indicators, And satisfy The reason for setting the weights in this way is that, under drought stress, net photosynthetic rate is a core indicator reflecting the plant's photosynthetic capacity and growth potential, which is directly related to yield and survival ability; therefore, it is given the highest weight. Relative water content of leaves is a key physiological parameter for assessing plant water status and drought resistance, and is crucial for maintaining cell turgor pressure and metabolic stability. Secondly, proline, as an osmotic regulator, can protect cell structure under drought conditions, but its effect is relatively indirect, hence its lower weight. Lower; fold change in gene expression Although it can reflect the strength of gene regulation, its physiological effects need to be reflected through downstream physiological indicators, therefore the weighting is... Further reduction; malondialdehyde, as a product of membrane lipid peroxidation, reflects the degree of oxidative damage in its absolute value, but because it is usually a negative indicator and has a limited range of variation, it is given the lowest weight. This ranking reflects the priority logic from direct physiological functions to indirect regulatory signals.

[0122] For this formula, the dependent variable Used to quantify the overall drought resistance strength of target genes under drought stress. The higher the value, the better the plant performs in terms of water retention, photosynthetic maintenance, osmotic regulation, oxidative damage mitigation, and gene expression regulation after the gene is overexpressed, that is, the more significant the drought resistance function of the gene. The smaller the value, the more limited or no significant positive effect the gene has on drought resistance.

[0123] This formula employs a linear weighted combination, unifying different physiological indicators and expression levels into a dimensionless index. Its clear structure and strong interpretability, along with the weights set according to differences in physiological importance, reflect the approach of multi-indicator synergistic evaluation. Taking the absolute value and assigning a negative sign expresses its negative effect as a damage indicator; for Taking the logarithm avoids numerical mutations caused by excessively high expression levels, which is more consistent with the biological principle that dose-response relationships often exhibit a logarithmic response.

[0124] Table 1: Statistics of Comprehensive Functional Verification Index

[0125]

[0126] Analysis of the data in Table 1 shows that there are differences in the relative change rate and expression fold of leaf relative water content, net photosynthetic rate, malondialdehyde content, proline content, and leaf expression fold among different groups. The fluctuations of these indicators jointly affect the trend of the comprehensive functional verification index. Overall, as the relevant physiological indicators adjust in a direction that is conducive to drought resistance, the expression fold changes accordingly, and the comprehensive functional verification index also shows a gradual upward trend.

[0127] The differences in the comprehensive functional validation index directly reflect the drought resistance potential of target genes in different groups. The higher the comprehensive functional validation index, the stronger the synergistic adaptation between physiological indicators and gene expression levels, which means that the corresponding target gene is more likely to have significant drought resistance. This data pattern provides an important reference for subsequent screening of potato genes with efficient drought resistance and also confirms the rationality of comprehensively evaluating gene drought resistance through multiple indicators.

[0128] The calculated comprehensive functional verification index Compared with the preset third threshold If a comparison is made, If the target drought-resistant gene is identified as a potato drought-resistant gene with significant drought-resistant function, then the target drought-resistant gene is determined to be a potato drought-resistant gene.

[0129] Wherein, the third threshold The determination method is as follows: Based on wild-type plant populations with the same genetic background, under the same drought stress conditions as in step S1, obtain relative change rate data of key physiological indicators and corresponding gene expression level change fold data of multiple groups of wild-type plants; calculate the corresponding comprehensive functional verification index for each group of data according to the formula; and set a critical value reflecting a significant improvement in drought resistance function through statistical methods or based on experience as the third threshold. .

[0130] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0131] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0132] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0133] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for identifying drought-resistant genes in potatoes using hyperspectral imaging, characterized in that, include: Set up drought stress group and control group for potato plants, and simultaneously acquire hyperspectral image data of the canopy and transcriptome expression profile data of each group of plants to construct an associated dataset; A spectral feature set including differential vegetation index, photochemical reflectance index, water stress index and red edge position parameter was extracted from hyperspectral image data, and a candidate gene expression level dataset in response to drought stress was screened from transcriptome expression profile data. Based on Spearman rank correlation analysis, the correlation coefficient between each feature in the spectral feature set and the expression level of each candidate gene in the candidate gene expression dataset is calculated. For each candidate gene, the absolute values ​​of its correlation coefficients with all spectral features are sorted in descending order. Candidate genes are then selected and included in the initial gene set based on the sorting. Each drought-resistant plant was randomly divided into a training set and a validation set. The spectral feature set in the training set was used as the feature input, and the expression level of each gene in the initial gene set was used as the feature output to train the random forest regression model. The spectral features of the validation set were input into the trained model to obtain the predicted expression level of each gene. The determination coefficient between the predicted expression level and the actual expression level of each gene was calculated. Based on the determination coefficient, the target drought-resistant genes were selected. Using the target drought-resistant gene as the object, transgenic plants overexpressing it were constructed. The relative change rate of key physiological indicators of transgenic plants relative to wild-type plants under drought stress was measured. Combined with the fold change of the expression level of the target drought-resistant gene in the transgenic plants, it was determined whether the gene is a potato drought-resistant gene with significant drought resistance function.

2. The method for identifying potato drought resistance genes using hyperspectral imaging according to claim 1, characterized in that: The process of establishing the associated dataset is as follows: Potato plants from the same batch, at the same growth stage, and with consistent physiological states were selected. One plant was randomly selected as the normal water supply control group and maintained with normal water supply. The remaining plants were treated as the drought stress group. At the time point when the experimental treatment reached the predetermined stress level, the reflectance spectrum images of the canopy of each group of living plants were simultaneously acquired using a hyperspectral imager under fixed light conditions. At the same time, within one hour after the spectral acquisition, fresh leaf samples from the same physiological part of each group of plants were taken, and RNA was immediately extracted and sequenced to obtain their transcriptome expression profile data. Each plant's hyperspectral image data is uniquely associated with its own transcriptome expression profile data to form a one-to-one associated dataset.

3. The method for identifying potato drought resistance genes using hyperspectral imaging according to claim 2, characterized in that: The hyperspectral image data is preprocessed by radiometric calibration, atmospheric correction and geometric correction in sequence. For the preprocessed hyperspectral image data, the average reflectance spectral curve of the effective imaging area of ​​the canopy of each plant is extracted, and the difference vegetation index, photochemical reflectance index, water stress index and red edge position parameter are calculated based on the spectral curve to form a spectral feature set. The formula for calculating the difference vegetation index is as follows: ； In the formula, The difference is the vegetation index; The average reflectance in the near-infrared band; The average reflectance of the red band; The formula for calculating the photochemical reflectance index is as follows: ； In the formula, Photochemical reflectance index; , They are respectively At wavelength and Reflectivity at wavelength; The formula for calculating the water stress index is as follows: ； In the formula, The water stress index; The average reflectance of the shortwave infrared band 1; The red-edge position parameter is calculated by measuring the spectral curve. The wavelength corresponding to the maximum value of the first derivative within the red-edge region is determined; Simultaneously, from the transcriptome expression profile data, the expression levels of each gene in each drought-stressed plant in the drought stress group and the normal water supply control group are compared in pairs. The fold change in the expression level of each gene between the two groups is calculated. Genes with an absolute fold change in expression level greater than a preset fold threshold are selected as candidate genes. The candidate gene expression level data are then compiled to form a candidate gene expression level dataset.

4. The method for identifying potato drought resistance genes using hyperspectral imaging according to claim 1, characterized in that: Based on Spearman rank correlation analysis, the correlation coefficient between each feature in the spectral feature set and the expression level of each candidate gene in the candidate gene expression dataset is calculated. The specific logic is as follows: For the first in the candidate gene expression level dataset The candidate gene, and the first candidate gene in the spectral feature set. Using spectral features, the expression level sequence of this gene on all drought-stricken plants in the associated dataset was obtained, respectively. and the sequence of characteristic values ​​of this spectral feature on the corresponding plant. ,in This is an index of candidate genes in the candidate gene expression level dataset. This serves as an index for spectral features, including the difference vegetation index, photochemical reflectance index, water stress index, and red edge position parameter. This represents the number of drought-stricken plants in the associated dataset. An index for drought-affected plants; Based on expression level sequences Construct expression level sequence Based on eigenvalue sequences Constructing eigenvalue level sequences ,in, for In sequence Ascending order ranking in the middle for In sequence The values ​​are ranked in ascending order, and if there are ties, the average ranking value is used.

5. A method for identifying potato drought resistance genes using hyperspectral imaging according to claim 4, characterized in that: The correlation coefficient between each spectral feature in the spectral feature set and the expression level of each candidate gene in the candidate gene expression dataset is calculated using the following formula: ； ； In the formula, Indicates the first The candidate gene and the first Spearman's rank correlation coefficient for each spectral feature; Indicates the first On the first drought-stricken plant The expression levels of each candidate gene are related to the first... The difference in the grade of each spectral feature value; For the For each candidate gene, obtain the set of absolute values ​​of its correlation coefficients with all spectral features. Sort the values ​​in this set in descending order, and select the top [values] from the sorted sequence. The absolute value of each correlation coefficient is used to calculate this. arithmetic mean of the values ,Will With preset threshold If a comparison is made, Then the first One candidate gene was included in the initial gene set.

6. The method for identifying potato drought resistance genes using hyperspectral imaging according to claim 5, characterized in that: The expression level of each gene in the initial gene set is used as the regression prediction target, and all spectral feature values ​​in the spectral feature set are used as input features to construct a random forest regression model. First, each drought-stricken plant is randomly divided into a training set and a validation set. The training set is used to train the random forest regression model, and the validation set is used to evaluate the predictive performance of the model. In the training set, each spectral feature value in the spectral feature set is standardized to obtain standardized spectral feature values. The expression level of each gene in the initial gene set is used as the corresponding regression output target and input into the random forest regression model for training. Using a pre-trained random forest regression model, for each plant individual in the validation set, its corresponding standardized spectral feature value is input to obtain the predicted expression level of each gene in the validation set samples. For each gene, the coefficient of determination between its predicted expression level and its actual expression level in the validation set samples is calculated using the following formula: ； In the formula, The coefficient of determination; Indicates the initial selection of genes in the gene pool. In the verification set The true expression level of each plant. To verify the index of the concentrated plants. This serves as an index for the genes in the initial gene pool. Indicates the initial selection of genes in the gene pool. In the verification set Predicted expression levels for each plant; Indicates the initial selection of genes in the gene pool. The average value of the true expression level in the validation set; To verify the number of plants collected; The prediction error for each gene in the initial gene set is calculated using the following formula: ； in, Indicates the initial selection of genes in the gene pool. The prediction error; The calculated coefficient of determination With the preset first threshold Compare and set the prediction error With the preset second threshold Compare, if satisfied and Then the initially selected gene The target drought-resistant gene was identified.

7. The method for identifying potato drought resistance genes using hyperspectral imaging according to claim 6, characterized in that: For each target drought-resistant gene, a corresponding overexpression transgenic potato plant was constructed, and wild-type plants with the same genetic background were used as a control. Both the transgenic plant group and the wild-type plant group were subjected to drought stress treatment consistent with the conditions set in step 1. After the treatment reached the predetermined stress level, the key physiological indicators of each plant were measured. The key physiological indicators included relative leaf water content, net photosynthetic rate, malondialdehyde content and proline content. The relative rate of change of each key physiological indicator between the transgenic plant group and the wild-type plant group was calculated. At the same time, the expression level of the target drought resistance gene in each plant was measured, and the fold change of the expression level of the target drought resistance gene in the transgenic plant relative to the wild-type plant was calculated. Among them, for each key physiological indicator The formula for calculating its relative rate of change is as follows: ； In the formula, Indicates the first The relative rate of change of key physiological indicators It serves as an index for key physiological indicators. For the first The arithmetic mean of the measured values ​​of several key physiological indicators in all plants in the transgenic plant group. For the first The arithmetic mean of the values ​​of the key physiological indicators measured in all plants in the wild-type plant group; The formula used to calculate the fold change in expression level is as follows: ； In the formula, The change in expression level is a factor of 5. The expression level of the target drought-resistant gene is the arithmetic mean of the expression levels of all plants in the transgenic plant group. The expression level of the target drought-resistant gene is the arithmetic mean of the expression levels of all plants in the wild-type plant group; Based on the relative rate of change and the fold change in expression level mentioned above, the comprehensive functional validation index is determined using the following formula: ； In the formula, The comprehensive functional verification index; , , , These represent the relative changes in leaf relative water content, net photosynthetic rate, malondialdehyde content, and proline content, respectively. , , , , Assign a preset weight to the corresponding indicator, and satisfy the following conditions: ; The calculated comprehensive functional verification index Compared with the preset third threshold If a comparison is made, If the target drought-resistant gene is identified as a potato drought-resistant gene with significant drought-resistant function, then the target drought-resistant gene is determined to be a potato drought-resistant gene.

Images