A method for detecting resistance to powdery mildew in rosa based on beta-alanine
By using hyperspectral imaging technology and machine learning algorithms to detect the β-alanine content in rose leaves, the accuracy problem of rose powdery mildew resistance detection has been solved, enabling rapid and accurate resistance assessment and promoting the intelligent development of agriculture.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- JIANGSU UNIV
- Filing Date
- 2026-04-14
- Publication Date
- 2026-07-21
AI Technical Summary
Existing methods for detecting rose powdery mildew resistance are not accurate enough and are greatly affected by rose variety variability and changes in environmental conditions, making it difficult to quickly and accurately determine the disease resistance of rose plants.
Hyperspectral imaging technology was used to measure the β-alanine content in rose leaves. An analytical model was established by combining it with machine learning algorithms. By measuring the relationship between β-alanine content and powdery mildew, a resistance grading model was established, and different levels of resistance were output.
It enables rapid and accurate detection of powdery mildew resistance in roses, reduces labor and time costs, improves detection accuracy, allows for early disease detection, reduces pesticide use, and promotes the intelligent development of agriculture.
Smart Images
Figure CN122430469A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the interdisciplinary field of agricultural information technology and plant pathology, and specifically relates to a method for detecting powdery mildew resistance in roses based on β-alanine. Background Technology
[0002] Rose ( Rosa chinensis Jacq. is a member of the genus Rosa in the family Rosaceae (Juss.). Rosa Roses are perennial plants, blooming year-round with beautiful flowers, and are widely distributed as ornamental plants in many cities both domestically and internationally, earning them the title of "Queen of Flowers." However, roses are susceptible to various pathogens, including the obligate parasitic fungus *Monoclonus roseus* (*Rosa monofilariae*). Sphaerotheca pannosa Powdery mildew, caused by [unspecified disease], is the leading cause of damage to cut rose production worldwide. It can infect leaves, buds, and petioles, severely impacting the ornamental quality of roses. Therefore, rapid identification of powdery mildew resistance in rose plants is crucial for improving the economic benefits for rose growers.
[0003] Currently, morphological characteristics are used to assess powdery mildew resistance in roses. It is generally believed that rose varieties susceptible to powdery mildew have large, thin leaves with few hairs or a thin waxy layer, and red and aromatic roses generally exhibit weaker resistance. However, in actual production, the pathogen causing rose powdery mildew is highly variable, leading to a sharp decline in the resistance of rose varieties, transforming them from highly resistant to moderately resistant, or even susceptible. Therefore, a new method for detecting powdery mildew resistance in roses is urgently needed for testing the resistance of existing varieties.
[0004] Existing research indicates that β-alanine in rose buds is essential for the germination of powdery mildew conidia, and the higher the β-alanine content in rose leaves, the higher the incidence of powdery mildew. Based on this, this invention aims to quantify the resistance of roses to powdery mildew by using β-alanine content in rose buds using hyperspectral technology. The invention utilizes non-destructive methods such as hyperspectral technology to obtain the β-alanine content in rose leaves. Hyperspectral technology is a non-destructive testing technique capable of acquiring the reflectance or emissivity of an object's surface across multiple continuous narrow wavelength bands. It can capture the spectral characteristics of an object in different wavelength bands, thereby enabling precise identification of the object's material, structure, and chemical composition. This invention patent employs hyperspectral imaging technology, enabling rapid detection of rose disease resistance without damaging the rose plant, improving detection accuracy, reducing labor and time costs, allowing for early disease detection, reducing pesticide use, and promoting the intelligent development of agriculture. Summary of the Invention
[0005] To address the aforementioned technical problems, this invention provides a method for detecting powdery mildew resistance in roses based on β-alanine, thereby resolving the issues in the prior art. The technical solution adopted by this invention is as follows: A method for detecting powdery mildew resistance in roses based on β-alanine includes: Step 1: Measure the content of β-alanine in rose leaves; Step 2: Establish an analytical model between β-alanine and powdery mildew in roses; Step 3: Establish a resistance grading model and output different levels of resistance.
[0006] Furthermore, in step 1, the content of β-alanine in diseased and healthy rose leaves is measured using high-performance liquid chromatography, including: Total amino acids were extracted from rose leaves using a kit, and the content of β-alanine was then measured using liquid chromatography.
[0007] Furthermore, step 2 includes: measuring the β-alanine content in healthy and diseased leaves of rose varieties with different resistance, comparing the differences in β-alanine content between healthy and diseased, susceptible and highly resistant rose varieties, thereby establishing an analytical model between the β-alanine content in rose leaves and rose powdery mildew, and determining the β-alanine content threshold when rose varieties are infected with powdery mildew.
[0008] Furthermore, step 3 includes: extracting spectral reflectance characteristics of leaves from different varieties in healthy and diseased states, and combining these with texture features to enhance the expression of structural differences between varieties; using a feature selection algorithm to screen for the optimal band combination. A resistance grading model is established to output different levels of resistance, including: highly resistant, moderately resistant, and susceptible.
[0009] Furthermore, in step 3, hyperspectral data of healthy and diseased rose plants are acquired, and noise and baseline drift are eliminated. Then, the optimal band combination is screened using a continuous projection algorithm, a competitive adaptive reweighted sampling method, and a principal component analysis method.
[0010] This invention offers the following advantages: By measuring the β-alanine content in rose leaves and establishing an analytical model between β-alanine and rose powdery mildew, this invention can accurately quantify the resistance of roses to powdery mildew. Compared to traditional morphological methods, this invention is unaffected by factors such as rose variety variability and changes in environmental conditions, thus exhibiting higher accuracy and reliability. Attached Figure Description
[0011] Figure 1 This is a flowchart; Figure 2 The result is the first derivative of the hyperspectral data under healthy conditions. Figure 3The result is the first derivative of the hyperspectral data under disease conditions. Figure 4 This is a flowchart of the support vector machine operation. Detailed Implementation
[0012] The following will be illustrated with reference to the embodiments of the present invention. Figures 1-4 The technical solutions in the embodiments of the present invention will be clearly and completely described. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Unless otherwise specified, the technical means used in the embodiments are conventional means well known to those skilled in the art.
[0013] A method for detecting powdery mildew resistance in roses based on β-alanine includes: Step 1: Measure the β-alanine content in rose leaves; specifically: The content of β-alanine in diseased and healthy rose leaves was measured using high performance liquid chromatography (HPLC). (1) Total amino acids were extracted from rose leaves using a kit, and the content of β-alanine was measured using a liquid chromatograph. The liquid chromatography parameters were as follows: mobile phase A was 0.02 mol / L NaAc-Hac, filtered through a 0.45 μm aqueous membrane; mobile phase B was acetonitrile, which increased from 5% to 40% within 0-25 min, filtered through a 0.45 μm aqueous membrane. (2) A C18 column (5 μm, 4.6 mm × 250 mm) was selected, with a flow rate of 1.0 ml / min, a column temperature of 30 °C, a detection wavelength of 360 nm, a reference wavelength of 600 nm, and an injection volume of 20 μL.
[0014] Step 2: Establish an analytical model between β-alanine and powdery mildew in roses; The β-alanine content in the leaves of different rose varieties varies. Susceptible rose varieties have higher β-alanine content in their leaves than resistant varieties, which is related to the fact that β-alanine promotes the reproduction of rose powdery mildew fungus. By measuring the β-alanine content in healthy and infected leaves of rose varieties with different resistance levels, and comparing the differences in β-alanine content between healthy, infected, and highly resistant susceptible rose varieties, an analytical model was established between the β-alanine content in rose leaves and rose powdery mildew, thus determining the threshold for β-alanine content when rose varieties are infected with powdery mildew.
[0015] Specific procedure: The β-alanine content of leaves from different rose varieties was measured, and then reflectance spectra of leaves from different rose varieties were collected using a ground-based spectrometer. First, the reflectance spectral data were preprocessed using centralization (CT), Savitzky-Golay (SG), standardization (SS), de-trending (DT), moving average (MA), and maximum minimum standardization (MMS). Then, principal component analysis was used to reduce the dimensionality of the spectral data. Finally, a predictive model for β-alanine content was established using support vector regression (SVR), k-nearest neighbor (KNN), and partial least squares (PLS) methods.
[0016] Step 3: Establish a resistance grading model and output different levels of resistance; Spectral reflectance characteristics of leaves from different varieties under healthy and diseased conditions were extracted (the β-alanine content of rose leaves was used in this experiment), and texture features (GLCM, Gabor filtering, etc.) were combined to enhance the expression of structural differences among varieties. Feature selection algorithms (such as SPA, CARS, RF importance ranking) were used to screen the optimal band combination. A resistance grading model (such as SVM, RandomForest, XGBoost) was established to output levels such as "highly resistant", "moderately resistant", and "susceptible". Specific steps: 1. Obtain hyperspectral data of healthy and diseased rose plants; 2. Use SG filtering, detrending, standard normal transformation (SNV), derivative transformation (first and second derivatives) to eliminate noise and baseline drift.
[0017] The derivation formula of the SG filtering algorithm is as follows: (1) Let the width of the filter window be y = 2x + 1, and the measurement points be n = x. i A k-1 degree polynomial is used to fit the data points within the window: (2) The formula for calculating the residual is: (3) The least squares solution A is: (4) The largest predicted or filtered value of the Z model is: Standard Normal Transform (SNV) Correction Formula: Where i = 1, 2, 3, ..., n; j = 1, 2, 3, ..., m; r ij denoted as spectral reflectance of the sample, n as the sample size, and m as the number of bands.
[0018] The specific formula for the first derivative is: The specific formula for the second derivative is: 3. SPA (Continuous Projection Algorithm), CARS (Competitive Adaptive Reweighted Sampling), and PCA (Principal Component Analysis) are used to screen the optimal band combination. Specifically: Competitive Adaptive Reweighted Sampling (CARS): This algorithm divides the dataset into training and test sets in a 4:1 ratio to construct a PLS model. The modeling process first requires setting the number of sampling iterations T and then calculating the absolute value weights of the regression coefficients in the PLS model, where |x i |and w i Let represent the absolute value of the regression coefficient of the i-th variable and the weight of the absolute value of the regression coefficient, respectively, and m be the number of variables remaining after each sampling.
[0019] The weight calculation formula is as follows: In the i-th modeling process, according to the scaling parameter E i The wavelength to be retained is determined.
[0020] E i The calculation formula is: Where μ and k are constants, their calculation formula is: Continuous Projection Algorithm (SPA): Let the number of rose samples be m, the number of bands be n, its spectral matrix m×n be X, and N be the number of feature wavelengths to be extracted. The SPA feature extraction calculation process is as follows: (1) Let the spectral matrix X m×n column vector x i For x k(0) Then the set of positions of the remaining row vectors is: (2) Calculate the projection Px of column vector xi onto the remaining column vectors step by step. i : (3)Select the maximum projection vector and extract the corresponding characteristic wavelength: (4)Set the maximum projection vector as the starting point of the next projection cycle: (5)Set u = u + 1. If u < N, then return to (2) for recalculation and loop in this way until u = N, then the characteristic wavelength is obtained as x k(n) , {n = 1, …, N - 1}.
[0021] Principal component analysis (PCA), the formula is as follows: In the formula, PC i is the score of the first i principal components, Z is the standardized matrix of the original variables, A i is the eigenvector corresponding to the i-th eigenvalue when the eigenvalues of the correlation coefficient matrix of the original variables are arranged in descending order, PC is the comprehensive score of the principal components, is the n-th eigenvalue when the cumulative contribution rate reaches 85%. ) 4. Model selection: Select the optimal classification model using SVM (Support Vector Machine), Random Forest, and BP neural network methods.
[0022] The specific process is as follows: Sample division: Stratified sampling is carried out according to variety and resistance level to ensure that the distributions of the training set, validation set, and test set are consistent.
[0023] Label definition: Define the objective criteria for "high resistance", "medium resistance", and "susceptible" (such as the range of disease index DI).
[0024] Model training and parameter tuning: Use grid search (Grid Search) or Bayesian optimization (Bayesian Optimization) to find the optimal hyperparameters.
[0025] Model evaluation: This method constructs a model based on machine learning and the content of β-alanine in roses, which can be widely applied to the open-air and greenhouse rose cultivation environments, achieving early warning of rose powdery mildew, predicting the disease situation in advance, and making timely response measures to reduce the impact of powdery mildew on roses. In addition, it can also be used for the screening of rose resistance breeding. Just scan the spectrum of the leaves of the new rose variety and import it into the model to judge whether the new variety is a powdery mildew-resistant variety. This method is more efficient, convenient, and non-destructive than using molecular biology methods to detect the resistance of new rose varieties. Determine the most suitable grading index to evaluate the established model and determine the optimal model.
[0026] The specific process of this invention is as follows: Reflectance spectra of leaves from different rose varieties in both healthy and diseased states are collected. First, the collected reflectance spectral data are preprocessed using methods including centralization (CT), Savitzky-Golay (SG), standardization (SS), de-trending (DT), moving average (MA), and maximum minimum standardization (MMS). Principal component analysis is used to reduce the dimensionality of the spectral data. Then, a classification model is built using a backpropagation neural network, support vector machine, and random forest algorithms. The prediction performance of the classification models is then compared, with the highest Rp2 value being the objective to determine the optimal classification model. This method incorporates β-alanine, a physiological indicator highly correlated with the target plant and the disease, into traditional machine learning modeling. The model constructed by combining these two factors is more targeted at rose powdery mildew, resulting in more accurate results.
[0027] The above embodiments are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Any modifications, alterations, alterations, or substitutions made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.
Claims
1. A method for detecting powdery mildew resistance in roses based on β-alanine, characterized in that, include: Step 1: Measure the content of β-alanine in rose leaves; Step 2: Establish an analytical model between β-alanine and powdery mildew in roses; Step 3: Establish a resistance grading model and output different levels of resistance.
2. The method for detecting powdery mildew resistance in roses based on β-alanine according to claim 1, characterized in that, In step 1, high-performance liquid chromatography (HPLC) was used to measure the β-alanine content in diseased and healthy rose leaves, including: Total amino acids were extracted from rose leaves using a kit, and the content of β-alanine was then measured using liquid chromatography.
3. The method for detecting powdery mildew resistance in roses based on β-alanine according to claim 1, characterized in that, Step 2 includes: measuring the β-alanine content in healthy and diseased leaves of rose varieties with different resistance, comparing the differences in β-alanine content between healthy and diseased, susceptible and highly resistant rose varieties, and establishing an analytical model between β-alanine content in rose leaves and rose powdery mildew, and determining the threshold of β-alanine content when rose varieties are infected with powdery mildew.
4. The method for detecting powdery mildew resistance in roses based on β-alanine according to claim 1, characterized in that, Step 3 includes: extracting spectral reflectance characteristics of leaves from different varieties in healthy and diseased states, and combining texture characteristics to enhance the expression of structural differences among varieties; using a feature selection algorithm to screen the optimal band combination. A resistance grading model is established to output different levels of resistance, including: highly resistant, moderately resistant, and susceptible.
5. The method for detecting powdery mildew resistance in roses based on β-alanine according to claim 4, characterized in that, In step 3, hyperspectral data of healthy and diseased rose plants are acquired, and noise and baseline drift are eliminated. Then, the optimal band combination is selected using the continuous projection algorithm, the competitive adaptive reweighted sampling method, and the principal component analysis method.