WO2022080411A1 - Method for predicting soybean yield - Google Patents

Abstract

Provided is a method for predicting soybean yield at an early stage with high accuracy. A method for predicting soybean yield, comprising acquiring analysis data of one or more components from a leaf sample collected from soybean, and predicting the soybean yield by using a correlation between the data and the soybean yield.

Description

Rice yield prediction method

The present invention relates to a method for predicting the yield of rice at an early stage.

Rice is an important grain and is widely eaten in Japan and around the world. In addition, the production volume in Japan is higher than that of other typical grains such as corn, wheat and soybean. It is widely cultivated as one of such important grains, and techniques for increasing the yield are being developed.
The growing period of rice varies slightly depending on the variety and cultivation conditions, but it usually takes a long period of 3 to 6 months from sowing to harvesting. Therefore, in the development of a technique for increasing the yield of rice, it takes a lot of time for cultivation to evaluate the yield. Furthermore, under seasonal and climatic conditions such as Japan, rice cultivation, which takes 3 to 6 months to harvest, is generally cultivated once a year. Since the yield evaluation in outdoor cultivation can be performed only once a year, which is an obstacle to the development of the yield increasing technique, a method for predicting the yield at an early stage has been required. In addition, if the yield can be predicted at an early stage in the actual production situation, the producer can easily determine whether or not to introduce a costly additional technique in order to secure a stable yield.

Various methods for early evaluation of yield using meteorological data, field images and growth information have been studied so far. For example, in Non-Patent Document 1, a model for predicting production at the national level is constructed using meteorological data such as cumulative precipitation and cumulative average temperature before sowing. Further, in Patent Document 1 and Patent Document 2, an attempt is made to measure the amount of nitrogen in the leaf blade, the color of the leaf, or the amount of chlorophyll, evaluate the growth and yield, and determine the amount of fertilizer applied. Furthermore, in Non-Patent Document 2, metabolites extracted from rice seeds or from above-ground parts about 15 days after sowing are comprehensively measured by GC-MS, and a hybrid rice yield prediction model is used using these data. It is reported that it was created.

However, the model of Non-Patent Document 1 is a model for predicting the yield of rice using meteorological data predicted before sowing, but the predictable unit is the national unit, and the predictors and yields for each individual are used. It is not suitable for evaluation when you want to correspond. The methods of Patent Documents 1 and 2 can be said to be non-destructive and simple measurement, but the prediction time is after the panicle formation stage, that is, after half of the growth period has elapsed. Furthermore, since the prediction is made on a field-by-field basis, it is not a technique for predicting the yield at the individual level. Further, in Non-Patent Document 2, the predictability evaluation of a model called cross-validation, which is performed when constructing a normal predictive model, is not performed. Specifically, although R ² indicating the prediction accuracy of the model is reported to be 0.82, there is no description of Q ² indicating the predictability of the important model together with R ² . There was concern that the model of Non-Patent Document 2 is an overfitted model in which the error for model construction data is small but the prediction error for unknown data is large. Further, the method of Non-Patent Document 2 is invasive and is not suitable for evaluation when it is desired to correspond the predictor of each individual with the yield.

[Patent Document 1] Japanese Patent Application Laid-Open No. 2000-300077 [Patent Document 2] Japanese Patent Application Laid-Open No. 2018-82648 [Non-Patent Document 1] Iizumi, T. et al., Climate Services, 2018, vol. 11, p.13 -twenty three
[Non-Patent Document 2] Dan, Z. et al., Scientific Reports, 2016, 6, 21732

The present invention is a rice yield prediction method, which obtains analysis data of one or more components from a leaf sample collected from rice and predicts the rice yield by using the correlation between the data and the rice yield. offer.

The figure which shows the relationship between the predicted value and the actually measured value of the yield by the OPLS model constructed using the data of Non-Patent Document 2. The figure which shows the relationship between the predicted value and the actually measured value of the yield by the OPLS model constructed using all 26 data. Analytical data of all components from the 11th place to the 800th place, analysis data of all the components from the 21st place to the 800th place, and analysis data of all the components from the 31st place to the 800th place in the model of FIG. ^R2 (indicated as R2Y in the figure) and ^Q2 (indicated as Q2 in the figure) value of each model constructed by the OPLS method using the analysis data of all the components from the 111th place to the 800th place. The figure which shows. Of the analysis data of the components from the top 1 to 10 VIP values in the model of FIG. 2, R ² (R2Y in the figure) of each model constructed by the OPLS method for any 9 combinations (10 combinations). The figure which shows the value (display) value and ^Q2 (displayed as Q2 in the figure) value.

Detailed description of the invention

The present invention relates to providing a method for accurately predicting the yield of rice at an early stage.

As a result of various studies on the evaluation of the yield of rice, the present inventors have found that the metabolites contained in the leaves have a component whose abundance correlates with the yield, and that one of the developed leaves of rice is used at an early stage after sowing. It was found that the final yield can be evaluated at the individual level by collecting the parts, analyzing the components contained in the leaves, and analyzing the components.

According to the method of the present invention, the yield of rice can be predicted at an early stage. As a result, for example, it becomes easy to determine the introduction of additional techniques for ensuring the yield, and it is possible to greatly improve the efficiency of the development of the techniques for increasing the yield.

In the present invention, rice means all plants belonging to the genus Oryza (scientific name Oryza) in the family Gramineae, and preferably means Asian rice (scientific name Oryza sativa) belonging to the genus Oryza in the family Gramineae. Asian rice includes Japonica and Indica. The varieties included in the Japonica species are Nihonbare, Milky Queen, Koshihikari, Hitomebore, Hinohikari, Akitakomachi, Nanatsuboshi, Haenuki, Masigura, Kinuhikari, Asahi no Yume, Yumepirika, Kinumusume, Koshihikari, Tsuyahime, and Yume Tsuyahime. , Fusakogane, etc. Examples of the varieties included in the indica species include Mudgo, Peta, Yodanya, Sigadis, IR8, IR36, IR72 and the like. As described above, there are a wide variety of varieties, but the present invention is not limited thereto.

The growth stages from the emergence of rice to the heading stage are the germination stage (around 5 days after sowing), the two-leaf age stage (around 25 days after sowing), the tillering stage (around 60 days after sowing), and the panicle formation stage (around 60 days after sowing). It is divided into a tiller stage (around 100 days after sowing) and a heading stage (around 120 days after sowing). In the present invention, the rice leaf sample may be collected from the germination stage to the heading stage when the leaves can be collected, and the collection time is preferably from the two-leaf instar stage to the heading stage, more preferably. The two-leaf instar to the stop leaf stage, more preferably the two-leaf instar to the panicle formation stage, and further preferably the two-leaf instar to the splitting stage. It is preferable that the number of days before and after each of the above growth stages is within 5 days.
Alternatively, the time for collecting rice leaves is 5 days or more after sowing, preferably 16 days or more, more preferably 25 days or more, still more preferably 40 days or more, and preferably 90 days or more after sowing, more preferably. It can be before 70 days, more preferably before 60 days. Further, it may be 5 to 90 days after sowing, preferably 16 to 70 days, more preferably 25 to 60 days, and even more preferably 40 to 60 days.
Here, the number of days after sowing represents the number of days when seedlings are transplanted and cultivated outdoors. In direct sowing cultivation in an indoor house, the number of days after sowing of each growth stage is different from the above number of days, but those skilled in the art can understand the number of days after sowing of each growth stage in consideration of the direct sowing cultivation conditions in the indoor house. For example, the tillering period is about 60 days after sowing when transplanted outdoors, but about 30 days after sowing when directly sown in an indoor house. In direct sowing cultivation in an indoor house, leaf samples may be collected at a time corresponding to the above-mentioned collection time.

The site for collecting the leaf sample is not particularly limited, but for example, about 1 to 5 leaves, preferably about 2 to 6, and more preferably about 4 to 7 leaves are cut from the root of the strain and collected. Be done.

In the present invention, the yield means the actual amount of rice seedlings per individual or the number of grains per individual. The mass is not particularly limited, but a dry mass is preferable. Of these, the actual amount of dried rice per individual is particularly preferable. The actual amount of dried child in the present invention means the mass measured in a state where the water content contained in the grain after drying at 90 ° C. for 72 hours is reduced to 7% or less. It is preferable that the actual amount of the desiccant is measured only by a balance with a calibration function such as an electronic balance.

In the present invention, the analysis data of the components to be obtained include high performance liquid chromatography (HPLC), gas chromatography (GC), ion chromatography, mass spectrometry (MS), near infrared spectrometric analysis (NIR), and Fourier conversion. Infrared spectroscopic analysis (FT-IR), nuclear magnetic resonance analysis (NMR), Fourier transform nuclear magnetic resonance analysis (FT-NMR), inductively coupled plasma mass spectrometer (ICP-MS), liquid chromatograph and mass spectrometry. Examples thereof include data analyzed / measured using a combined LC / MS or other instrument analysis means, preferably mass spectrometric data, and more preferably mass spectrometric data by LC / MS.
Examples of the mass spectrometric data include precision mass (“m / z value”), ionic strength, holding time, and the like, but precision mass information is preferable.

In order to apply the leaf sample to the above-mentioned instrumental analysis means, it is appropriately pretreated according to the analysis means, but usually, the collected leaves are wrapped in aluminum foil and immediately frozen in liquid nitrogen to stop the metabolic reaction. After being freeze-dried and dried, it is subjected to an extraction operation.
Extraction is performed by pulverizing the freeze-dried leaf sample using a bead crusher or the like, adding an extraction solvent, and stirring the mixture. Examples of the extraction solvent used here include methanol, ethanol, butanol, acetonitrile, chloroform, ethyl acetate, hexane, acetone, isopropanol, water and the like, and a mixture thereof. When LC / MS is used as the analytical means, an 80v / v% aqueous methanol solution to which an internal standard substance is added is preferably used.

In the present invention, examples of the components in the leaves to be analyzed include rice metabolites separated and detected by LC / MS. Preferably, the components provided by mass spectrometry have a precise mass (m / z) of 101-1215. More preferably, the 1,324 components listed in Tables 1a to 1i below, which are defined by the precise mass (m / z value) provided by mass spectrometry, can be mentioned. In the process of separation detection by LC / MS, if different molecular ion peaks of partial decomposition products and adducts (M + H, M + Na, etc.) are generated from the metabolites, the detected partial decomposition products are different from the original metabolites. It was used as a component of.

The 1,324 components are selectively extracted from the metabolites of rice, and the selection method is as shown in the examples in detail. Directly sown and cultivated, 2) 4 to 7 leaves were collected about 1 month after sowing to obtain leaf samples, 3) components were extracted using 80v / v% methanol, and then 4) LC / MS analysis. To obtain molecular ion information (precision mass, m / z) and structural information derived from fragments, 5) extract peaks derived from components, and then align each peak between each sample, isotopes. The analysis data of 1,324 components is acquired by removing peaks, correcting peak intensity between samples, and removing noise. The method of correcting the peak intensity between the samples is not particularly limited, and examples thereof include correction using the popled QC method. In the popled QC method, a sample called poored QC is prepared by mixing a certain amount from all the samples in the same batch, and the popled QC is analyzed at a constant frequency (about once every 5 to 9 times) between each sample. By implementing, the estimated value "what will happen to each peak intensity if it is assumed that the QC sample was analyzed when each sample was analyzed" is calculated, and the process of correcting with that value is performed. It corrects the sensitivity between each sample. The data correction method does not significantly affect the correlation with yield and the performance of the prediction model.

In addition, the result of correlation analysis between the acquired analysis data of 1,324 components and the corresponding yield data (dried material real amount) (single correlation coefficient r between the peak area and yield of the analysis data of each component and none). The p-value was calculated by the correlation test), and it was shown that certain components significantly correlated with the yield (see Tables 4a-4q below).

Therefore, among the 1,324 components, the component to be analyzed in the present invention is a component having a significant correlation with the yield (p <0.05) and an absolute value of the correlation coefficient | r |> 0.51, that is, Ingredient No. Selected from 10, 177, 178, 245, 254, 272, 294, 337, 366, 435, 462, 259, 359, 708, 729, 832, 842, 869, 901, 912, 1050, 1050, 1173 and 1306. It is preferable to include one or more kinds. All of the above components had VIP values of 1.08 or more, which will be described later.

Further, among the 1,324 components, the component to be analyzed in the present invention is a component having a significant correlation with the yield (p <0.05) and an absolute value of the correlation coefficient | r |> 0.66, that is, a component. No. It is preferable to include one or more selected from 10, 178 and 1173. All of the above components had VIP values of 2.17 or more, which will be described later.

In Tables 1a to 1i, the components of 1,324 are defined by the precise mass obtained by mass spectrometry, and the composition formula of the compound can be estimated from these precise mass data. Further, the partial structure information of the compound can be obtained from the MS / MS data acquired at the same time as the analysis. Therefore, the target component can be estimated from the composition formula and the partial structure information, and the one that can be compared with the reagent can be identified.

For example, as a result of the analysis, No. 10 is the composition formula C ₆ H ₉ N ₃ , No. 178 is the composition formula C ₁₃ H ₁₈ O ₃ , No. 245 is the composition formula C ₁₃ H ₂₀ O ₄ , and No. 272 is the composition. Formulas C ₁₃ H ₂₈ O ₄ , No. 347 are composition formulas C ₁₈ H ₂₆ O ₂ , No. 416 and No. 417 are composition formulas C ₁₈ H ₂₈ O ₃ , No. 539 is composition formula C ₁₅ H ₂₂ O ₈ , No. 729 is the composition formula C ₁₉ H ₃₀ O ₇ , and No. 1050 is the composition formula C ₂₄ H ₂₈ O ₁₁ , No. 1182 was presumed to have the composition formula C ₃₄ H ₄₀ O ₁₀ .

As a means for predicting the yield of rice, the above 1,324 components, preferably components having a significant correlation with the yield (p <0.05) and an absolute value of the correlation coefficient | r |> 0.51. Of the precision mass m / z 124.0869 with a preferably significant (p <0.05) and absolute value of the correlation coefficient | r |> 0.66 (eg, a correlation coefficient of −0.825). (Peak area) can also be measured for the rice leaf sample to be predicted, and the yield can be predicted from the correlation between the known yield and the measured peak area.

In addition, the yield can be predicted by using a plurality of the analysis data of the above 1,324 components and collating with the yield prediction model constructed by using the multivariate analysis method.
That is, a rice leaf sample after a lapse of a predetermined period from sowing is collected, an analysis sample is obtained, the analysis sample is subjected to instrumental analysis to obtain instrumental analysis data, and the instrumental analysis data is collated with a yield prediction model. Therefore, the yield of the rice can be predicted.

The yield prediction model can be constructed by performing regression analysis using the peak area value of the corrected component analysis data with each precise mass as the explanatory variable and the yield value as the objective variable. Regression analysis methods include, for example, principal component regression analysis, PLS (Partial least squares projection to latent structures) regression analysis, OPLS (Orthogonal projections to latent structures) regression analysis, generalized linear regression analysis, bagging, and support vector machines. Examples include multivariate regression analysis methods such as machine learning / regression analysis methods such as random forest and neural network regression analysis. Of these, it is preferable to use the PLS method, the OPLS method which is an improved version of the PLS method, or the machine learning / regression analysis method. The OPLS method has the same predictability as the PLS method, but is superior in that it is easier to visualize for interpretation for the purpose of this time. Both the PLS method and the OPLS method are methods in which information is aggregated from high-dimensional data, replaced with a small number of latent variables, and the objective variable is expressed using the latent variables. It is important to properly select the number of latent variables, and cross-validation is often used to determine the number of latent variables. That is, the data for model construction is divided into several groups, one group is used for model verification, and the other group is used for model construction to estimate the prediction error, and this work is repeated while exchanging the groups to obtain the prediction error. The number of latent variables with the smallest total is chosen.

The evaluation of the prediction model is mainly judged by two indicators. One is R ² which represents prediction accuracy, and the other is Q ² which represents predictability. R ² is the square of the correlation coefficient between the measured value of the data used for constructing the prediction model and the predicted value calculated by the model, and the closer it is to 1, the higher the prediction accuracy. On the other hand, ^Q2 is the result of the above cross validation, and represents the square of the correlation coefficient between the actually measured value and the predicted value which is the result of repeated model validation. In the rice yield prediction model of the present invention, it is preferable to use ^Q2 > 0.50 as the standard for model evaluation. Since R ² > Q ² is always satisfied, Q ² > 0.50 satisfies R ² > 0.50 at the same time.
The following shows the results of creating various rice yield prediction models using the peak area values of the analysis data of the above 1,324 components and the grain yield and verifying their accuracy.

(1) Construction of a yield prediction model using all component information An OPLS model was constructed from a total of 26 data matrices having peak area values and yield values of analysis data of 1,324 components per data. At the time of construction, the peak area value and the yield data of the analysis data of each component were converted into an average of 0 and a variance of 1 by autoscaling.
In the above model, the contribution to the model performance given to each component called the VIP (Variable Impact in the Projection) value is calculated.
The VIP value is obtained by the following formula 1.

The larger the VIP value, the greater the contribution to the model, and it also correlates with the absolute value of the correlation coefficient. A list of the top 800 VIP values is shown in Tables 5a to 5j below.

(2) Model construction using the VIP value as an index (2-1) Model using analysis data of the components up to the top 800 VIP values Select all the components up to the top 800 VIP values and select the 800 per data. An OPLS model (Fig. 2) was constructed from a total of 26 data matrices having peak area values and yield values of the analysis data of the individual components. At the time of construction, the peak area value and the yield data of the analysis data of each component were converted into an average of 0 and a variance of 1 by autoscaling. R ² = 0.78 and Q ² = 0.51, which means that the model has high predictability.

(2-2) Model using analysis data of components with lower VIP values among the components up to the top 800 VIP values Analysis data of all components from the 11th place to the 800th place with the VIP value, up to the 800th place from the 21st place A model (Fig. 3) was created by the OPLS method using the analysis data of all the components of the above, the analysis data of all the components from the 31st position to the 800th position, and the analysis data of all the components from the 111th position to the 800th position. It was constructed.
^Q2 > 0.5 is satisfied by a model using the analysis data of all the components from the 11th position to the 800th position and the analysis data of all the components from the 21st position to the 800th position. Even if the analysis data of all the components from the VIP value 31st to the 800th is used, ^Q2 > 0.50 does not hold.

(2-3) Model using 9 analysis data of the components from the top 10 VIP values Arbitrary 9 combinations (10 ways) of the analysis data of the components from the top 1 to 10 VIP values A model (Fig. 4) was constructed by the OPLS method.
Both models satisfy ^Q2 > 0.50.

The number of components used for prediction is preferably as small as possible, for example, 10 or less, preferably 5 or less, more preferably 3 or less, and most preferably 3 or less, in the case of simple prediction. It is one. Further, when it is desired to improve the accuracy, it is preferable that the number of components is large, for example, 11 or more, preferably 20 or more, more preferably 50 or more, still more preferably 90 or more, and most preferably 150. More than one. When predicting with a small number of components, it is preferable to use a component having a higher VIP value or a component having a higher correlation coefficient for prediction.

The component having a higher VIP value is, for example, at least one component selected from the top 800 VIP values, preferably at least five components selected from the top 800 VIP values, and more preferably a VIP value. At least 10 components selected from the top 800, more preferably at least 1 component selected from the top 10 VIP values, and even more preferably at least 5 selected from the top 10 VIP values. It is a component of, more preferably at least 9 components selected from the top 10 VIP values, and even more preferably a component with the top 800 VIP values.

The embodiments and preferred embodiments of the present invention are shown below.
<1> From the leaf sample collected from rice, analysis data of one or more components selected from the components having a precise mass (m / z) of 101 to 1215 provided by mass spectrometry is acquired, and the data and the data are obtained. A rice yield prediction method that predicts rice yield using the correlation with rice yield.
<2> The method according to <1>, wherein the analysis data of the one or more components is corrected by the popled QC method.
<3> The component is at least one selected from the components shown in Tables 1a to 1i defined by the precision mass (m / z) provided by mass spectrometry, <1> or <2>. The method described in.
<4> The components are the component Nos. No. 1 shown in Tables 1a to 1i. 1, 4, 6, 9, 10, 11, 12, 19, 20, 21, 23, 26, 27, 29, 30, 33, 34, 35, 38, 39, 45, 46, 47, 48, 49, 50, 51, 52, 54, 55, 56, 61, 62, 63, 64, 65, 66, 67, 69, 71, 75, 76, 77, 78, 81, 83, 84, 85, 88, 89, 90, 91, 92, 96, 100, 102, 105, 106, 107, 108, 109, 113, 116, 118, 119, 120, 121, 123, 124, 126, 127, 129, 130, 131, 133, 134, 137, 139, 142, 145, 147, 148, 149, 151, 152, 153, 154, 155, 156, 159, 162, 163, 164, 165, 166, 167, 168, 169, 170, 172, 174,175,177,178,181,182,183,184,186,187,188,191,193,194,196,198,200,202,203,206,208,209,210,212,213, 214, 215, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, 229, 230, 231, 232, 233, 235, 237, 238, 239, 240, 243, 245, 246, 247, 248, 249, 250, 251, 252, 254, 255, 258, 259, 260, 261 and 262, 263, 264, 265, 266, 268, 270, 271, 272, 273, 274, 275, 277, 278, 280, 281, 283, 284, 285, 286, 288, 289, 290, 291, 293, 294, 298, 300, 302, 303, 305, 310, 312, 314, 316, 317, 318, 320, 321, 323, 325, 327, 329, 331, 332, 333, 334, 335, 337, 338, 339, 342, 343, 344, 345, 346, 347, 348, 350, 351 and 352, 355, 358, 359, 360, 361, 362, 363, 365, 366, 368, 369, 370, 371, 373, 374, 375, 378, 379, 381, 382, 389, 390, 391, 392, 395, 397, 398, 399, 401, 404, 407, 408, 409, 410, 411, 413, 414, 415, 416, 41 7, 418, 419, 423, 424, 425, 428, 431, 433, 434, 435, 436, 437, 438, 439, 441, 444, 445, 446, 447, 449, 450, 451, 454, 455, 457, 458, 459, 460, 461, 462, 464, 465, 469, 471, 472, 473, 474, 475, 478, 480, 481, 482, 483, 487, 489, 490, 491, 492, 494, 502, 503, 504, 507, 509, 510, 511, 512, 513, 514, 516, 517, 522, 523, 525, 526, 259, 532, 534, 359, 540, 542, 543, 547, 548, 549, 551, 552, 554, 555, 557, 561, 565, 566, 567, 573, 582, 583, 585, 586, 588, 589, 590, 591, 593, 594, 595, 596, 599, 599, 600, 602, 603, 604, 606, 609, 611, 612, 613, 615, 616, 617, 618, 620, 621, 624, 628, 630, 631, 632, 633, 635, 639, 643, 644, 647, 649, 650, 651, 653, 654, 655, 656, 658, 660, 661, 662, 665, 666, 671, 672, 673, 674, 675, 681, 682, 683, 684, 685, 688, 689, 691, 692, 693, 694, 695, 696, 699, 700, 701, 702, 703, 704, 706, 707, 708, 713, 714, 715, 717, 719, 721, 722, 723, 724, 725,726,727,728,729,731,732,734,735,737,738,740,745,746,748,749,750,754,756,757,762,765,766,767,768, 770, 774, 767, 777, 780, 781, 782, 785, 787, 789, 792, 793, 794, 795, 796, 797, 798, 799, 801, 802, 803, 804, 810, 811, 813, 815, 816, 817, 818, 820, 822, 823, 824, 827, 828, 829, 830, 832, 834, 841, 842, 843, 844, 845, 846, 848, 849, 850, 852, 854, 85 8,863,864,867,868,869,870,871,872,874,877,878,879,882,883,884,885,886,888,889,893,894,895,896,898, 899, 900, 901, 902, 903, 910, 911, 912, 914, 917, 919, 922, 923, 924, 925, 926, 928, 930, 923, 938, 941, 942, 943, 944, 945, 946, 947, 948, 949, 950, 952, 953, 955, 956, 958, 959, 960, 962, 965, 966, 968, 969, 973, 976, 979, 980, 981, 983, 985, 986, 989, 992, 993, 994, 995, 996, 997, 999, 1001, 1002, 1003, 1005, 1006, 1007, 1009, 1012, 1013, 1015, 1017, 1019, 1020, 1021, 1022, 1024, 1025, 1026, 1027, 1031, 1032, 1034, 1036, 1039, 1043, 1044, 1045, 1046, 1047, 1048, 1049, 1050, 1051, 1053, 1054, 1057, 1058, 1059, 1060, 1062, 1066, 1067, 1068, 1069, 1070, 1072, 1074, 1075, 1077, 1078, 1079, 1081, 1082, 1087, 1088, 1089, 1092, 1094, 1098, 1100, 1101, 1102, 1103, 1104, 1105, 1106, 1108, 1110, 1112, 1113, 1114, 1117, 1118, 1119, 1120, 1121, 1123, 1126, 1127, 1128, 1129, 1133, 1134, 1135, 1139, 1140, 1141, 1142, 1143, 1144, 1147, 1148, 1149, 1150, 1151, 1152, 1153, 1154, 1158, 1160, 1163, 1166, 1167, 1168, 1170, 1171, 1172, 1173, 1174, 1177, 1178, 1179, 1180, 1181, 1182, 1184, 1186, 1187, 1188, 1189, 1190, 1191, 1192, 1193, 1194, 1195, 1196, 1197, 1198, 1199, 1202, 1204, 1208, 1211, 1212, 1214, 1217, 1218, 1221, 1222, 1224, 1225, 1226, 1229, 1231, 1233, 1234, 1235, 1237, 1238, 1239, 1240, 1241, 1242, 1243, 1244, 1246, 1247, 1248, 1249, 1250, 1252, 1254, 1255, 1256, 1257, 1258, 1261, 1263, 1265, 1267, 1268, 1269, 1271, 1272, 1276, 1277, 1278, 1280, 1283, 1291, 1292, 1295, 1296, Described in <3>, which is one or more selected from 1297, 1299, 1300, 1301, 1304, 1305, 1306, 1309, 1311, 1312, 1313, 1314, 1315, 1316, 1317, 1318, 1319, 1321 and 1322. the method of.
<5> The components are the component Nos. No. 1 shown in Tables 1a to 1i. Selected from 10, 177, 178, 245, 254, 272, 294, 337, 366, 435, 462, 259, 359, 708, 729, 832, 842, 869, 901, 912, 1050, 1050, 1173 and 1306. The method according to <3>, which is one or more types.
<6> The components are the component Nos. No. 1 shown in Tables 1a to 1i. 10. The method according to <3>, which is one or more selected from 10, 178 and 1173.
<7> The components are the component Nos. No. 1 shown in Tables 1a to 1i. It is one or more selected from 10, 178, 245, 272, 347, 416, 417, 359, 729, 1050 and 1182, and the component No. 10 is a component of the composition formula C ₆ H ₉ N ₃ and described above. Component No. 178 is a component of composition formula C ₁₃ H ₁₈ O ₃ , said component No. 245 is a component of composition formula C ₁₃ H ₂₀ O ₄ , and component No. 272 is a component of composition formula C ₁₃ H ₂₈ O. The component _No. 347 is a component of the composition formula C ₁₈ H ₂₆ O ₂ , the component No. 416 is a component of the composition formula C ₁₈ H ₂₈ O ₃ , and the component No. 417 is a component of the composition formula C 18 H 28 O 3. It is a component of the composition formula C ₁₈ H ₂₈ O ₃ , and the component No. 539 is a component of the composition formula C ₁₅ H ₂₂ O ₈ , and the component No. 729 is a component of the composition formula C ₁₉ H ₃₀ O ₇ , and the component No. 1050 is a component of the composition formula C ₂₄ H ₂₈ O ₁₁ . The method according to <3>, wherein 1182 is a component of the composition formula C ₃₄ H ₄₀ O ₁₀ .
<8> The method according to any one of <1> to <7>, wherein the leaf sample is collected from rice in the germination stage to the heading stage.
<9> The method according to any one of <1> to <7>, wherein the leaf sample is collected from rice in the two-leaf instar to the panicle formation stage.
<10> The method according to any one of <1> to <9>, wherein the analysis data is mass spectrometry data.
<11> At least one of the top 800 VIP values calculated from the yield prediction model constructed using the component information shown in Tables 1a to 1i of the component analysis data obtained from the leaf sample. The method according to any one of <3> to <10>, which comprises a step of collating with a yield prediction model constructed using the analysis data of the components of.
<12> In <11>, the yield prediction model uses at least 5 of the top 800 VIP values calculated from the yield prediction model constructed using the component information shown in Tables 1a to 1i. The method described.
<13> In <11>, the yield prediction model uses at least 10 of the top 800 VIP values calculated from the yield prediction model constructed using the component information shown in Tables 1a to 1i. The method described.
<14> In <11>, the yield prediction model uses at least one of the top 10 VIP values calculated from the yield prediction model constructed using the component information shown in Tables 1a to 1i. The method described.
<15> In <11>, the yield prediction model uses at least 5 of the top 10 VIP values calculated from the yield prediction model constructed using the component information shown in Tables 1a to 1i. The method described.
<16> In <11>, the yield prediction model uses at least 9 out of the top 10 VIP values calculated from the yield prediction model constructed using the component information shown in Tables 1a to 1i. The method described.
<17> The method according to <11>, wherein the yield prediction model uses the top 800 VIP values calculated from the yield prediction model constructed by using the component information shown in Tables 1a to 1i.
<18> The method according to any one of <11> to <17>, wherein the yield prediction model is a model constructed by using the OPLS method.
<19> The method according to any one of <1> to <18>, wherein the precision mass is measured with an accuracy of four or more digits after the decimal point.

Comparative example 1. Acquisition of data for analysis We obtained the data (https://www.nature.com/articles/srep21732#Sec8) published together with the literature of Dan et al. Of Non-Patent Document 2. The actual amount of dried child per individual was used as the yield data. For leaf extract data, all published data was used for analysis.

2. 2. Model construction / evaluation A multivariate analysis method was used to construct a yield prediction model using analysis data of two or more components, and SIMCA ver. 14 (Umetrics) was used. For the prediction model, regression analysis was performed using the peak area value of the corrected component analysis data with each precise mass as the explanatory variable and the yield value as the objective variable. Regression analysis was performed by the OPLS method, which is an improved version of the PLS method.

The evaluation method of the prediction model is mainly judged by two indexes. One is R ² which represents prediction accuracy, and the other is Q ² which represents predictability. R ² is the square of the correlation coefficient between the measured value of the data used for constructing the prediction model and the predicted value calculated by the model, and the closer it is to 1, the higher the prediction accuracy. On the other hand, Q ² is the result of the cross validation, and represents the square of the correlation coefficient between the actually measured value and the predicted value which is the result of repeated model validation. As in the examples, ^Q2 > 0.50 was used as the criterion for model evaluation. Since R ² > Q ² is always satisfied, Q ² > 0.50 satisfies R ² > 0.50 at the same time.

3. 3. Construction / evaluation of a model using all data An OPLS model was constructed from a total of 295 data matrices, which has peak area values and yield values of analysis data of 525 components per data. At the time of construction, the peak area value and the yield data of the analysis data of each component were converted into an average of 0 and a variance of 1 by autoscaling. As a result of model construction, R ² = 0.07 indicating prediction accuracy and Q ² = 0.008 indicating predictability did not meet the criteria of Q ² > 0.50. The results are shown in FIG. Therefore, it was found that the yield prediction model of Non-Patent Document 2 has very low prediction accuracy.

Example 1. Cultivation test The data of the pot cultivation test in the greenhouse conducted in 2019 will be described in detail.
Pot cultivation was carried out in a greenhouse in Hiratsuka City, Kanagawa Prefecture. The soil used was the field soil in the Tochigi Plant of Kao Corporation. The basic fertilizer application amount was set to 0.8 g per pot, and a chemical fertilizer containing nitrogen, phosphorus and potassium as fertilizer components (trade name "Hyakukatsu Ichiki" Kanryo Chemical Co., Ltd.) was mixed with 4 L of soil. By setting the conditions of 1/4 times, 1/2 times, 2 times and 4 times the basic fertilizer application amount, cultivation was carried out under a total of 5 types of fertilizer application conditions. A 1 / 5000a Wagner pot was used as a pot, and about 4 L of the above soil was packed per pot to prepare 30 pots. On July 29, 2019, 3 seeds were sown at 2 places in each pot (6 seeds were used per pot). Plants derived from three seeds were collectively treated as one individual. The cultivar used was Japonica rice "Nihonbare". When two true leaves were unfolded, they were thinned out so that there would be one plant per pot. From August 22nd to October 28th, the plants were cultivated under flooded conditions, and from sowing to the start of flooding, watering was carried out once a week to the extent that the soil was moistened. No watering was done after October 28th. Sampling was done on August 28th. After sampling, 1 g of chemical fertilizer containing 14% each of nitrogen, phosphoric acid, and potassium (trade name "Chemical Fertilizer No. 14" Sun and Hope Co., Ltd.) was topped up. Harvesting was carried out on 14th November (80 days after sowing). Since 4 individuals were deficient, a total of 26 individuals were used for yield prediction. The temperature inside the greenhouse was adjusted appropriately by opening and closing the door according to the temperature.

2. 2. Leaf sampling Leaf sampling was performed during the daytime, 30 days after sowing (generally from 13:00 to 15:00). The growth stage of rice at this time was slightly different depending on the individual, but the number of leaves per individual was about 15 to 20, which was a growth stage corresponding to the tillering stage. Leaf sampling was taken by cutting 4-7 leaves from the root of the plant. At the time of collection, the whole strain was collected without bias. The collected leaves were wrapped in aluminum foil and immediately frozen in liquid nitrogen to stop the metabolic reaction. The frozen sample was taken back to the laboratory while maintaining the frozen state, and dried by freeze-drying. This dried sample was subjected to the extraction operation described later.

3. 3. Final measurement of grain yield Harvesting was carried out on November 14, 80 days after sowing. All grains were collected from each individual after the cultivation test and dried in a dryer set at 90 ° C. (blower constant temperature incubator DKN6022, Yamato Kagaku Co., Ltd.) for 3 days. As the yield data, the actual amount of dried sardines (mgDW / individual) and the number of grains (pieces / individual) were measured. The dry cell parenchyma (mgDW / individual) was used for the simple correlation analysis between the analysis data of each component and the yield described later and the construction of the prediction model. As shown in Table 2, the data on the parenchymal amount of dried child was 5507.97 mgDW / individual at the minimum and 10507.17 mgDW / individual at the maximum.

4. Extraction of collected leaf components Freeze-dried leaf samples were manually crushed as much as possible using a spatula. After pulverization, 10 mg was weighed in a 2 mL tube (Safelock tube, Eppendorf), one zirconia ball having a diameter of 5 mm was added to the tube, and the mixture was pulverized at 25 Hz for 1 minute with a bead crusher (MM400, Resch). As the extraction solvent, an 80v / v% methanol aqueous solution containing lidocaine (Wako Pure Chemical Industries, Ltd., # 120-02671) at a concentration of 500 ng / mL was used as an internal standard. 1 mL of the prepared extraction solvent was added to the crushed tube, and homogenized extraction was performed at 20 Hz for 5 minutes using the same bead crusher. After the extraction is completed, the mixture is centrifuged in a desktop centrifuge (Chibitan) of about 2,000 × g for about 30 seconds, and filtered with a 0.45 μm hydrophilic PTFE filter (DISMIC-13HP 0.45 μm syringe filter, ADVANTEC). An analysis sample was obtained.

5. Analysis of leaf samples by LC / MS For analysis of leaf extraction samples, LC / MS analysis is performed using an Agilent HPLC system (Infinity 1260 series) as a front and an AB SCIEX Q-TOFMS device (TripleTOF4600) as a detector. I did it. Separation columns in HPLC include a core shell column Capcell core C18 (2.1 mm ID × 100 mm, particle total 2.7 μm) manufactured by Shiseido Co., Ltd. and a guard column (2.1 mm ID × 5 mm, particles). A total of 2.7 μm) was used, and the column temperature was set to 40 ° C. The autosampler was kept at 5 ° C during the analysis. The analytical sample was injected with 5 μL. A: 0.1 v / v% formic acid aqueous solution and B: 0.1 v / v% formic acid acetonitrile solution were used as eluents. The gradient elution condition was maintained at 1v / v% B (99v / v% A) for 0 to 0.1 minutes, and 1v / v% B to 99.5v / v for 0.1 to 13 minutes. The ratio of eluent B was increased to% B and maintained at 99.5 v / v% B from 13.01 minutes to 16 minutes. The flow velocity was 0.5 mL / min.

As for the mass spectrometer conditions, the ionization mode was set to the positive mode, and ESI was used as the ionization method. In this analysis system, the elution ions are scanned by TOFMS for 0.1 seconds, 10 high-intensity ions are selected, and each of them is subjected to MS / MS for 0.05 seconds while repeating the cycle. Molecular ion information (precision mass, m / z) obtained by TOFMS scanning and structural information derived from fragments generated by MS / MS scanning were acquired. The mass measurement range was set to m / z 100-1,250 for TOFMS and m / z 50-1,250 for MS / MS. The parameters of each scan are set to GS1 = 50, GS2 = 50, CUR = 25, TEM = 450, ISVF = 5500, DP = 80 and CE = 10 for TOFMS scans, and GS1 = for MS / MS scans. It was set to 50, GS2 = 50, CUR = 25, TEM = 450, ISVF = 5500, DP = 80, CE = 30, CES = 15, IRD = 30 and IRW = 15.

6. Creation of data matrix Data processing was performed as follows. First, peaks were extracted using MarkerView ^TM Software (AB SCIEX). The peak extraction condition (“peak finding option”) is a peak corresponding to a retention time of 0.5 to 16 minutes, 20 scans of Subtraction offset in the item of “Enhance Peak Finding”, 5 ppm of Minimum spectral peak width, and 5 ppm of Minimum spectral peak. .. Factor was set to 1.2, Minimum RT peak width was set to 10 scans, Noise threshold was set to 5, and the Assign charge state in the "More" item was checked. As a result, peak information of 31,649 was obtained.

Next, an alignment process was performed to align the detected peaks between the analyzed samples. As for the alignment processing conditions (“Alightment & Filtering”), the Retition time tradition in the item of “Alignment” was set to 0.20 minutes and the Mass tolerance was set to 10.0 ppm. In addition, the Integrity threshold in the "Filtering" item was set to 10, the Retention time filtering was checked, the Move peaks in <3 samples were set, and the Maximum number of peaks was set to 50,000. The retention time was corrected using the lidocaine peak in the item of "Internal standard".

Next, the isotope peak was removed. Since the isotope peak is automatically recognized by the software at the time of peak extraction and labeled as "isotopic" on the peak list, the corresponding peak was deleted by sorting by "isotopic". As a result, the peak decreased to 25,895 peaks.

Next, the peak intensity was corrected between the samples. In this analysis, in addition to the samples, a sample called poored QC was prepared by mixing a certain amount from all the samples, and the polled QC was analyzed once every 6 times. From all these QC analysis results, an estimated value "what will happen to each peak intensity if it is assumed that the QC sample was analyzed when each sample was analyzed" is calculated, and the process of correcting with that value is performed. This was done and the sensitivity between each sample in the same batch was corrected. For this treatment, free software (LOWESS-Normalization-Tool) provided by RIKEN was used. Finally, the 9 QC analysis data measured were used to calculate the relative standard deviation (RSD) of the 11,408 peaks, removing the highly variable peaks with RSD> 30% and finally 1,324. Peak data, i.e., analytical data for 1,324 components was obtained. The obtained analytical data are shown in Tables 3a to 3q. Subsequent analysis was performed using these data.

7. Correlation analysis Correlation analysis was performed using the analysis data of 1,324 components in the leaves of 26 individuals acquired and the corresponding yield data (real amount of dried matter), that is, the matrix data of 26 × 1,324. The p-value was calculated by the simple correlation coefficient r and the uncorrelated test between the analysis data of each component and the yield data. The results are shown in Tables 4a-4q. The "component No." in the table is for convenience, in which 1,324 components are numbered from the smallest mass number when arranged in order of mass. Further, the analysis result includes information on the holding time as well as the mass information. It has been shown that mass spectrometric data can be compared and analyzed between mass spectrometric samples. Therefore, the information on the holding time was removed, and only the precise mass information was described.

The results obtained by the correlation analysis showed that the components with a certain correlation coefficient significantly correlated with the yield. It was found that the absolute value of the correlation coefficient | r |> 0.51 was 24 components and | r |> 0.66 was 3 components.

8. Model construction / evaluation A multivariate analysis method was used to construct a yield prediction model using analysis data of two or more components, and SIMCA ver. 14 (Umetrics) was used. For the prediction model, regression analysis was performed using the peak area value of the corrected component analysis data with each precise mass as the explanatory variable and the yield value as the objective variable. Regression analysis was performed by the OPLS method, which is an improved version of the PLS method.

The evaluation method of the prediction model is mainly judged by two indexes. One is R ² which represents prediction accuracy, and the other is Q ² which represents predictability. R ² is the square of the correlation coefficient between the measured value of the data used for constructing the prediction model and the predicted value calculated by the model, and the closer it is to 1, the higher the prediction accuracy. On the other hand, Q ² is the result of the cross validation, and represents the square of the correlation coefficient between the actually measured value and the predicted value which is the result of repeated model validation. From a prediction point of view, the model is considered to have good predictability if at least ^Q2 > 0.50 (Triba, M.N. et al., Mol. BioSyst. 2015, 11, 13-19. .), ^Q2 > 0.50 was used as the standard for model evaluation. Since R ² > Q ² is always satisfied, Q ² > 0.50 satisfies R ² > 0.50 at the same time.

8-1. Construction / evaluation of a model using all data An OPLS model was constructed from a total of 26 data matrices, which has peak area values and yield values of analysis data of 1,324 components per data. At the time of construction, the peak area value and the yield data of the analysis data of each component were converted into an average of 0 and a variance of 1 by autoscaling. As a result of model construction, R ² = 0.931 indicating the prediction accuracy and Q ² = 0.344 indicating the predictability did not meet the criteria of Q ² > 0.50.

8-2. VIP value calculation In the model constructed in 8-1, VIP (Variable Impact in)
The degree of contribution to the model performance given to each component called the projection (variable importance in projection) value is calculated. The larger the VIP value, the greater the contribution to the model, and it also correlates with the absolute value of the correlation coefficient. A list of the top 800 VIP values is shown in Tables 5a to 5j.

8-3. Model construction using the VIP value as an index A model was constructed with a plurality of components based on the VIP value ranking (Tables 5a to 5j), which is the contribution of each component to the model constructed in 8-1. Although not particularly limited, the model performance standard is set to ^Q2 > 0.50 for convenience.

8-3-1. Model using the analysis data of the components up to the top 800 VIP values All the components up to the top 800 VIP values are selected, and each data has the peak area value and yield value of the analysis data of the 800 components. An OPLS model for predicting yield was constructed from a total of 26 data matrices. At the time of construction, the peak area value and the yield data of the analysis data of each component were converted into an average of 0 and a variance of 1 by autoscaling. As a result of model construction, R ² = 0.78 indicating prediction accuracy and Q ² = 0.51 indicating predictability. The results are shown in FIG. From this prediction model, it was shown that a model with high predictability can be constructed and early yield prediction is possible by using the component composition contained in the leaves for about one month of cultivation.

8-3-2. Model using analysis data of components with lower VIP values among the components up to the top 800 VIP values Analysis data of all components from the 11th place to the 800th place with VIP values, and all components from the 21st place to the 800th place An OPLS model was constructed using the analysis data, the analysis data of all the components from the 31st position to the 800th position, and the analysis data of all the components from the 111th position to the 800th position. As a result, ^Q2 > 0.5 is satisfied by the model using the analysis data of all the components from the 11th place to the 800th place and the analysis data of all the components from the 21st place to the 800th place, and the VIP value 31. It was found that ^Q2 > 0.50 does not hold even when the analysis data of all the components from the rank to the 800th rank are used (Fig. 3).

8-3-3. A model using 9 analysis data of the components of the top 10 VIP values Build an OPLS model for any 9 combinations (10 ways) of the analysis data of the components of the top 10 VIP values. I did it. As a result, it was found that ^Q2 > 0.50 was satisfied in all the models. From this, it was shown that a model with a certain predictability can be constructed if it contains 9 metabolites up to the top 10 VIP values (Fig. 4).

Claims (17)

1. From the leaf sample collected from rice, analysis data of one or more components selected from the components having a precise mass (m / z) of 101 to 1215 provided by mass spectrometry was obtained, and the data and the rice yield were obtained. A method for predicting the yield of rice, which predicts the yield of rice using the correlation of rice.
2. The method according to claim 1, wherein the analysis data of the above 1 or more components is corrected by the popled QC method.
3. The method according to claim 1 or 2, wherein the component is one or more selected from the components shown in Tables 1a to 1i below, which are defined by the precision mass (m / z) provided by mass spectrometry.
   

   

   
   

   

   
   

   

   
   

   

   
   

   

   
   

   

   
   

   

   
   

   

   
   

   
4. The components are the component Nos. No. 1 shown in Tables 1a to 1i. 1, 4, 6, 9, 10, 11, 12, 19, 20, 21, 23, 26, 27, 29, 30, 33, 34, 35, 38, 39, 45, 46, 47, 48, 49, 50, 51, 52, 54, 55, 56, 61, 62, 63, 64, 65, 66, 67, 69, 71, 75, 76, 77, 78, 81, 83, 84, 85, 88, 89, 90, 91, 92, 96, 100, 102, 105, 106, 107, 108, 109, 113, 116, 118, 119, 120, 121, 123, 124, 126, 127, 129, 130, 131, 133, 134, 137, 139, 142, 145, 147, 148, 149, 151, 152, 153, 154, 155, 156, 159, 162, 163, 164, 165, 166, 167, 168, 169, 170, 172, 174,175,177,178,181,182,183,184,186,187,188,191,193,194,196,198,200,202,203,206,208,209,210,212,213, 214, 215, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, 229, 230, 231, 232, 233, 235, 237, 238, 239, 240, 243, 245, 246, 247, 248, 249, 250, 251, 252, 254, 255, 258, 259, 260, 261 and 262, 263, 264, 265, 266, 268, 270, 271, 272, 273, 274, 275, 277, 278, 280, 281, 283, 284, 285, 286, 288, 289, 290, 291, 293, 294, 298, 300, 302, 303, 305, 310, 312, 314, 316, 317, 318, 320, 321, 323, 325, 327, 329, 331, 332, 333, 334, 335, 337, 338, 339, 342, 343, 344, 345, 346, 347, 348, 350, 351 and 352, 355, 358, 359, 360, 361, 362, 363, 365, 366, 368, 369, 370, 371, 373, 374, 375, 378, 379, 381, 382, 389, 390, 391, 392, 395, 397, 398, 399, 401, 404, 407, 408, 409, 410, 411, 413, 414, 415, 416, 41 7, 418, 419, 423, 424, 425, 428, 431, 433, 434, 435, 436, 437, 438, 439, 441, 444, 445, 446, 447, 449, 450, 451, 454, 455, 457, 458, 459, 460, 461, 462, 464, 465, 469, 471, 472, 473, 474, 475, 478, 480, 481, 482, 483, 487, 489, 490, 491, 492, 494, 502, 503, 504, 507, 509, 510, 511, 512, 513, 514, 516, 517, 522, 523, 525, 526, 259, 532, 534, 359, 540, 542, 543, 547, 548, 549, 551, 552, 554, 555, 557, 561, 565, 566, 567, 573, 582, 583, 585, 586, 588, 589, 590, 591, 593, 594, 595, 596, 599, 599, 600, 602, 603, 604, 606, 609, 611, 612, 613, 615, 616, 617, 618, 620, 621, 624, 628, 630, 631, 632, 633, 635, 639, 643, 644, 647, 649, 650, 651, 653, 654, 655, 656, 658, 660, 661, 662, 665, 666, 671, 672, 673, 674, 675, 681, 682, 683, 684, 685, 688, 689, 691, 692, 693, 694, 695, 696, 699, 700, 701, 702, 703, 704, 706, 707, 708, 713, 714, 715, 717, 719, 721, 722, 723, 724, 725,726,727,728,729,731,732,734,735,737,738,740,745,746,748,749,750,754,756,757,762,765,766,767,768, 770, 774, 767, 777, 780, 781, 782, 785, 787, 789, 792, 793, 794, 795, 796, 797, 798, 799, 801, 802, 803, 804, 810, 811, 813, 815, 816, 817, 818, 820, 822, 823, 824, 827, 828, 829, 830, 832, 834, 841, 842, 843, 844, 845, 846, 848, 849, 850, 852, 854, 85 8,863,864,867,868,869,870,871,872,874,877,878,879,882,883,884,885,886,888,889,893,894,895,896,898, 899, 900, 901, 902, 903, 910, 911, 912, 914, 917, 919, 922, 923, 924, 925, 926, 928, 930, 923, 938, 941, 942, 943, 944, 945, 946, 947, 948, 949, 950, 952, 953, 955, 956, 958, 959, 960, 962, 965, 966, 968, 969, 973, 976, 979, 980, 981, 983, 985, 986, 989, 992, 993, 994, 995, 996, 997, 999, 1001, 1002, 1003, 1005, 1006, 1007, 1009, 1012, 1013, 1015, 1017, 1019, 1020, 1021, 1022, 1024, 1025, 1026, 1027, 1031, 1032, 1034, 1036, 1039, 1043, 1044, 1045, 1046, 1047, 1048, 1049, 1050, 1051, 1053, 1054, 1057, 1058, 1059, 1060, 1062, 1066, 1067, 1068, 1069, 1070, 1072, 1074, 1075, 1077, 1078, 1079, 1081, 1082, 1087, 1088, 1089, 1092, 1094, 1098, 1100, 1101, 1102, 1103, 1104, 1105, 1106, 1108, 1110, 1112, 1113, 1114, 1117, 1118, 1119, 1120, 1121, 1123, 1126, 1127, 1128, 1129, 1133, 1134, 1135, 1139, 1140, 1141, 1142, 1143, 1144, 1147, 1148, 1149, 1150, 1151, 1152, 1153, 1154, 1158, 1160, 1163, 1166, 1167, 1168, 1170, 1171, 1172, 1173, 1174, 1177, 1178, 1179, 1180, 1181, 1182, 1184, 1186, 1187, 1188, 1189, 1190, 1191, 1192, 1193, 1194, 1195, 1196, 1197, 1198, 1199, 1202, 1204, 1208, 1211, 1212, 1214, 1217, 1218, 1221, 1222, 1224, 1225, 1226, 1229, 1231, 1233, 1234, 1235, 1237, 1238, 1239, 1240, 1241, 1242, 1243, 1244, 1246, 1247, 1248, 1249, 1250, 1252, 1254, 1255, 1256, 1257, 1258, 1261, 1263, 1265, 1267, 1268, 1269, 1271, 1272, 1276, 1277, 1278, 1280, 1283, 1291, 1292, 1295, 1296, 1297, 1299, 1300, 1301, 1304, 1305, 1306, 1309, 1311, 1312, 1313, 1314, 1315, 1316, 1317, 1318, 1319, 1321 and 1322. the method of.
5. The components are the component Nos. No. 1 shown in Tables 1a to 1i. Selected from 10, 177, 178, 245, 254, 272, 294, 337, 366, 435, 462, 259, 359, 708, 729, 832, 842, 869, 901, 912, 1050, 1050, 1173 and 1306. The method according to claim 3, wherein the method is one or more.
6. The components are the component Nos. No. 1 shown in Tables 1a to 1i. 10. The method according to claim 3, wherein the method is one or more selected from 10, 178 and 1173.
7. The method according to any one of claims 1 to 6, wherein the leaf sample is collected from rice from the germination stage to the heading stage.
8. The method according to any one of claims 1 to 6, wherein the leaf sample is collected from rice in the two-leaf instar to the panicle formation stage.
9. The method according to any one of claims 1 to 8, wherein the analysis data is mass spectrometry data.
10. The analysis data of the components obtained from the leaf sample is used for at least one component from the top 800 VIP values calculated from the yield prediction model constructed by using the component information shown in Tables 1a to 1i. The method according to any one of claims 3 to 9, comprising a step of collating with a yield prediction model constructed using analytical data.
11. The method according to claim 10, wherein the yield prediction model uses at least 5 out of the top 800 VIP values calculated from the yield prediction model constructed by using the component information shown in Tables 1a to 1i. ..
12. The method according to claim 10, wherein the yield prediction model uses at least 10 out of the top 800 VIP values calculated from the yield prediction model constructed by using the component information shown in Tables 1a to 1i. ..
13. The method according to claim 10, wherein the yield prediction model uses at least one of the top 10 VIP values calculated from the yield prediction model constructed by using the component information shown in Tables 1a to 1i. ..
14. The method according to claim 10, wherein the yield prediction model uses at least 5 out of the top 10 VIP values calculated from the yield prediction model constructed by using the component information shown in Tables 1a to 1i. ..
15. The method according to claim 10, wherein the yield prediction model uses at least 9 out of the top 10 VIP values calculated from the yield prediction model constructed by using the component information shown in Tables 1a to 1i. ..
16. The method according to claim 10, wherein the yield prediction model uses the top 800 VIP values calculated from the yield prediction model constructed by using the component information shown in Tables 1a to 1i.
17. The method according to any one of claims 10 to 16, wherein the yield prediction model is a model constructed by using the OPLS method.

Images