Method for determining the quality of a semen of a vertebrate animal

Abstract

A method is provided for determining quality of semen of a vertebrate animal. The method includes the following steps: measuring at least one absorption spectrum of a sample of the semen; selecting a number n of wave numbers .sub.j (j[1;n]) which are characteristic of the semen of the breed or of the species of the animal; determining from the absorption spectrum or spectra of a value of the absorption X.sub.j and/or a value of the second derivative of the absorption X.sub.j.sub.(j[1;n]) for each of the n wave numbers .sub.j (j[1;n]); and calculating a non-return rate Y at a predefined number of days from the absorption values X.sub.j and/or from the second derivative of the absorption X.sub.j previously determined.

Claims

1. A method for determining the quality of a semen of a vertebrate animal, the method comprising: measuring at least one absorption spectrum of a sample of said semen with a spectrometer; selecting a number n, with n4, of wave numbers .sub.j (j[1;n]) where the wave numbers are spatial frequencies in the spectrum characteristic of semen of a breed or of species of said animal; determining from said at least one absorption spectrum of at least one of a value of the absorption X.sub.j or a value of the second derivative of the absorption X.sub.j.sub.(j[1;n])for each of said n wave numbers .sub.j (j[1;n]); and calculating a non-return rate Y at a predefined number of days from at least one of said absorption values X.sub.j or the second derivative of the absorption X.sub.j previously determined.

2. The method for determining the quality of a semen according to claim 1 , wherein said non-return rate Y is calculated according to the mathematical law, wherein X.sub.j.sub.(j[1;n]) is the normalized second derivative of the absorption value for the wave number .sub.j and the weighting coefficients .sub.0 and .sub.j (j[1;n]) are constants.

3. The method for determining the quality of a semen according to claim 1, wherein the values of said weighting coefficients are obtained from a processing of measurements of the absorption spectra of a plurality of semen samples from a population of reference vertebrate, of which the non-return rates are known.

4. The method for determining the quality of a semen according to claim 1, wherein in said measuring, at least 2 absorption spectra of a sample of said seed are measured and said determining values of absorption or of second derivatives of the absorption comprises averaging said measured spectra from which said values of the absorption or the second derivatives of the absorption are determined.

5. The method for determining the quality of a semen according to claim 1, wherein the number n of wave numbers .sub.j (j[1;n]) is greater than or equal to 7.

6. The method for determining the quality of a semen according to claim 1, wherein said wave numbers are each representative of a molecule or set of molecules selected from the group consisting of: lipids; carbohydrates; proteins; nucleic acids; a combination of a lipid, carbohydrate, protein, or nucleic acid molecule with at least one other lipid, carbohydrate, protein, or nucleic acid molecule.

7. The method for determining the quality of a semen according to claim 1, wherein said vertebrate is a bull and the non-return rate is a 90-day non-return rate, said bull coming from a breed selected from the group consisting of the following breeds: Abondance, Barnaise, Bordelaise, Bretonne pie noir, Brune, Froment du Lon, Jersiaise, Montbliarde, Pie rouge des plaines, Prim'Holstein, Rouge flamande, Bleue du nord, Normande, Salers, Tarentaise.

8. The method for determining the quality of a semen according to claim 1, wherein at least one of said wave numbers .sub.j (j[1;n]) is selected in the wave number range [960 cm.sup.1; 1100 cm.sup.1], [1440 cm.sup.1; 1550 cm.sup.1] or [2800 cm.sup.1; 3200 cm.sup.1] or in the amide B band.

9. The method for determining the quality of a semen according to claim 1, wherein said wave numbers .sub.j (j[1;n]) are selected from the group consisting of 31651 cm.sup.1, 31361 cm.sup.1, 31031 cm.sup.1, 30711 cm.sup.1, 30261 cm.sup.1, 29771 cm.sup.1, 29751 cm.sup.1, 17361 cm.sup.1, 16891 cm.sup.1, 16611 cm.sup.1, 16181 cm.sup.1, 15151 cm.sup.1, 14481 cm.sup.1, 14301 cm.sup.1, 13441 cm.sup.1, 13381 cm.sup.1, 13361 cm.sup.1, 13161 cm.sup.1, 12141 cm.sup.1, 1183 1 cm.sup.1, 11031 cm.sup.1, 10991 cm.sup.1, 10201 cm.sup.1, 9751 cm.sup.1.

10. The method for determining the quality of a semen according to claim 9, wherein the number n of said selected wave numbers .sub.j (j[1;n]) is equal to thirteen and the values of said thirteen selected wave numbers .sub.j (j[1;n]) are: 31651 cm.sup.1, 31361cm.sup.1 ,31031 cm.sup.1, 29751 cm.sup.1, 17361 cm.sup.1, 16611 cm.sup.1, 15151 cm.sup.1, 14481 cm.sup.1, 14301 cm.sup.1, 13161 cm.sup.1, 12141 cm.sup.1, 10201 cm.sup.1, 9751 cm.sup.1.

11. The method for determining the quality of a semen according to claim 9, wherein the number n of said selected wave numbers .sub.j (j[1;n]) is equal to fifteen and the values of said fifteen selected wave numbers .sub.j (j[1;n]) are: 3071 1 cm.sup.1, 3026 1 cm.sup.1, 2977 1 cm.sup.1, 1689 1 cm.sup.1, 1618 1 cm.sup.1, 1515 1 cm.sup.1, 1430 1 cm.sup.1 , 1344 1 cm.sup.1, 1338 1 cm.sup.1, 1336 1 cm.sup.1, 1183 1 cm.sup.1, 1103 1 cm.sup.1, 1099 1 cm.sup.1, 1020 1 cm.sup.1, 975 1 cm.sup.1.

12. The method for determining the quality of a semen according to claim 1, wherein the measuring at least one absorption spectrum comprises preparing said sample from said semen.

13. The method for determining the quality of a semen according to claim 1, further comprising comparing said non-return rate Y with a predetermined threshold, for selecting said semen for reproduction purposes in case where said non-return rate Y is greater than or equal to said predetermined threshold.

14. The method for determining the quality of a semen according to claim 13, wherein the said threshold is greater than or equal to 0.4.

15. The method for determining the quality of a semen according to claim 1, wherein in said measuring, at least 3 absorption spectra of a sample of said seed are measured and said determining values of absorption or of second derivatives of the absorption comprises averaging said measured spectra from which said values of the absorption or the second derivatives of the absorption are determined.

16. The method for determining the quality of a semen according to claim 1, wherein the number n of wave numbers .sub.j (j[1;n]) is greater than or equal to 9.

17. The method for determining the quality of a semen according to claim 1, wherein the number n of wave numbers .sub.j (j[1;n]) is greater than or equal to 13.

18. The method for determining the quality of a semen according to claim 1, further comprising comparing said non-return rate Y with a predetermined threshold, for selecting said semen for reproduction purposes in case where said non-return rate Y is greater than or equal to said predetermined threshold, wherein the said threshold is greater than or equal to 0.4.

19. The method for determining the quality of a semen according to claim 1, further comprising comparing said non-return rate Y with a predetermined threshold, for selecting said semen for reproduction purposes in case where said non-return rate Y is greater than or equal to said predetermined threshold, wherein the said threshold is greater than or equal to 0.5.

20. The method for determining the quality of a semen according to claim 1, further comprising comparing said non-return rate Y with a predetermined threshold, for selecting said semen for reproduction purposes in case where said non-return rate Y is greater than or equal to said predetermined threshold, wherein the said threshold is greater than or equal to 0.6.

Description

5. LIST OF FIGURES

(1) Other features and advantages of the invention will become evident on reading the following description of one particular embodiment of the invention, given by way of illustrative and non-limiting example only, and with the appended drawings among which:

(2) FIG. 1 illustrates, in block diagram form, the steps of an exemplary embodiment of a method for determining the quality of a semen of a bull according to the invention;

(3) FIG. 2 represents MIR, derived and standardised spectra acquired from a straw, a supernatant and a pellet;

(4) FIG. 3 is a representation of the variance of a straw and of a pellet from the same ejaculate depending on the wave number;

(5) FIGS. 4A, 4B and 5 illustrate the correlation between NRR90 values calculated by a model with 20 wave numbers for ejaculate samples of Prim'Holstein bulls and known values for these same samples, respectively for 70 samples, for 16 samples and for all of these 86 samples; FIG. 6 is another representation of FIG. 5 in which two lines have been added, representing a difference in value of the NRR90 with respect to the model, respectively by +5 points and 5 points.

6. DETAILED DESCRIPTION OF THE INVENTION

(6) 6.1. Experimental Protocol

(7) The samples analysed are bovine ejaculates in the form of straws, preserved in liquid nitrogen. Preliminary analyses were performed on 130 ejaculates from 50 different bulls from the PrimHolstein breed. The NRR90 gross indicator was used to qualify the quality of ejaculates.

(8) 6.1.1. Sample Preparation

(9) During the sample preparation stage, in a first step, straws are thawed in a water bath at 37 C. for 30 seconds. In a second step, the straws are analysed. For this, the contents of the straws are placed in a 1.5 ml Eppendorf tube. Seven microliters are then deposited on the sensor for MIR spectral acquisition of a straw. In a third step, a supernatant is extracted by centrifugation at 3500 g for 5 min at 15 C. The supernatant is then deposited on the sensor for the acquisition of the supernatant spectrum. In a fourth step, the pellet is rinsed with 600 l of 0.9% NaCl and centrifugation is then performed. The pellet is again suspended in 3.5 l of 0.9% NaCl and deposited on the sensor for the acquisition of the pellet spectrum.

(10) 6.1.2. MIR Spectral Acquisition of a Straw

(11) The spectra are acquired in absorbance from 4000 to 400 cm.sup.1. The spectral resolution is set at 4 cm.sup.1, with a zero-filling factor of 2, and 64 scans are recorded.

(12) A sensor is placed in the spectrometer, the baseline is recorded to calibrate the device, then 7 l of sample is deposited on the sensor. The spectrum is recorded after 6 minutes.

(13) 6.1.3. Spectrum Processing

(14) The spectra are analysed in the 3800-940 cm.sup.1 domain, the absorption domain of the majority of biomolecules. A straight line is generated from 2800 to 1800 cm.sup.1 to eliminate the contribution of CO.sub.2, then the second derivative (Savisky-Golay algorithm with 13 smoothing points) is calculated from 3200 to 2800 cm.sup.1 and 1800 to 940 cm.sup.1. Then a vector normalisation of the second derivatives is carried out. Quality criteria are defined to reject nonconforming spectra.

(15) 6.1.4. Choice of the Matrix for the Prediction of the Quality of the Semen

(16) Three types of acquisitions were made and compared: ejaculate or total semen, pellet and supernatant (centrifugation of the semen). An observation of the spectra (see FIG. 2) highlights that the spectra of straw 210 and supernatant 220 have many similarities. Principal Component Analysis (PCA) is performed on the data in order to compare the spectra. It appears on the factorial map that the spectra acquired from the supernatant are very close to those acquired from the total ejaculate. Thus the spectra acquired from the straws contain a biochemical information very close to that of the seminal fluid. Or, the latter is strongly diluted (from 3 to 30 times) in a buffer before freezing and is therefore not determinant of the specific quality of the diluted/frozen/thawed semen of the bull since all the ejaculates are treated in the same way. Due to the greater or lesser dilution of the ejaculate, the MIR spectra of the straws necessarily reflect the biochemical composition of the dilution medium. This contribution to the MIR spectra may therefore mask that of the spermatozoa that are supposed to contain the spectral information that makes the difference from the point of view of fertility. Moreover, if a defect of fertility linked only to a lower quality of the seminal fluid (fructose, pH, . . . ) is considered, it is expected that this defect is offset by the dilution in the buffer medium. It turns out that the spectral information that makes the difference is to be found on the cells alone, rather than on the more or less diluted overall sample. The variability of the measurements between straw and pellet 230 was also taken into account. For this, the variances were calculated on the 3 spectra acquired for the same ejaculate. The raw spectra, that is the non-derivative spectra, were first standardised by the Multiplicative Scatter Correction MSC method in order to overcome the baseline variations. As shown in FIG. 3, with a representation of the variance as a function of wave number (in cm.sup.1) for a discussion of the straws 310 and the pellet 320, it appears generally that the signals acquired from the total semen have more variability especially in the range of 1000-940 cm.sup.1.

(17) The measurements are therefore carried out on the centrifugation pellet essentially containing the spermatozoa.

(18) 6.2. Construction of Models of Determining the Quality of a Semen

(19) 6.2.1. Reference Samples

(20) Eighty-six ejaculates, from 40 Holstein bulls, the 90-day gross non-return rate, or NRR90 gross is known, were used for the establishment of the law or equation to determine the quality of the ejaculate.

(21) These ejaculates come from bulls aged 11 months to 10.5 years at the time of collection of their ejaculates. The distribution of the number of ejaculates per bull is as follows: 8, 18 and 14 bulls respectively produced 1, 2 or 3 ejaculates.

(22) 6.2.2. Preparation of Straws:

(23) For each ejaculate, six straws are thawed by placing them in a water bath at 37 C. for 30 seconds.

(24) The contents of six straws are emptied into an Eppendorf tube 1.5 ml, then it undergoes a centrifugation at 3500 g for 5 min at 15 C. The supernatant is removed and the pellet is rinsed with 600 l of 0.9% NaCl. This step is renewed once. Following the second rinsing, a new centrifugation is applied and the pellet is again suspended in 10 l of 0.9% NaCl.

(25) 6.2.3. Acquisition of Spectra

(26) From this preparation, three spectra are acquired for each ejaculate. The acquisition of each of the spectra is carried out with a pitch of 2 cm.sup.1, on a spectrometer whose precision, whatever the position, that is to say the wave number, is 0.1 cm.sup.1.

(27) To acquire each of the spectra, 7 l of the suspension formed of the pellet and the saline solution are deposited in the sensor. The spermatozoa are irradiated in the mid-infrared and their absorption spectrum is recorded after 6 minutes.

(28) 6.2.4. Data Analysis

(29) The spectra are subject to quality control on various criteria before being selected for further analysis. Criteria considered for quality control include signal strength, interference, signal-to-noise ratio, and water content.

(30) The average of three spectra acquired is performed to obtain an averaged spectrum of the ejaculate.

(31) The averaged spectrum is processed according to the procedure previously described in paragraph 6.1.3.

(32) For all the samples, the spectra are divided into two categories, the calibration spectra from the analysis of 70 ejaculates and the validation spectra from the analysis of the remaining eighteen ejaculates. Calibration spectra are used to construct the model in order to relate a variable to explain, here the NRR90 gross, and explanatory variables that is to say an optimised selection of wave numbers of the spectrum. Once the model has been optimised, the validation spectra are used to evaluate the predictive performance of the model.

(33) The reduction of explanatory variables is performed by a genetic algorithm associated with a 10% cross-validation R-PLS. Once this reduction completed, the choice of variables is optimised by repeating 100 linear regressions (LRs), with a cross-validation on 10% of the initial population. A validation of the explanatory variables accepted is made by integrating them into the law or the linear equation for predicting unknown samples.

(34) 6.2.5. Construction of Models of Determining the Quality of a Semen

(35) The reference samples are split into a calibration subpopulation and a validation subpopulation. The learning process is carried out on th of the ejaculates, the validation on the 5th remaining and for each of the sub-populations, one maximises the number of ejaculates from different bulls while having a proportional representativity of the individuals in 3 classes of NRR90 defined as follows: NRR90<40%, 40%NRR9050% and NRR90>50%.

(36) 6.2.5.1. Mathematical Model

(37) The mathematical model used is defined by the formula
Y=.sub.0+.sub.j=1.sup.n.sub.jX.sub.j,

(38) wherein: Y is the NRR90 calculated for the ejaculate; n (n4) is the number of wave numbers .sub.j considered in the model; X.sub.j is the normalised second derivative of the value of the absorption to the wavenumbers .sub.j; .sub.0 is the offset at the origin; .sub.j (1jn) is the weighting coefficient of the value of the normalised second derivative of the absorption X.sub.j, delineated by its standard error.

(39) 6.2.5.2. Examples of Models

(40) Five models are constructed from respectively 7, 9, 13, 15 and 20 wave numbers, minimising the prediction error, RMSEP (Root-Mean-Square Error of Prediction).

(41) The spectral ranges of wavenumbers selected relate in particular the field of absorption of lipids 3200-2800 cm.sup.1, proteins (amide B band and domain 1440-1500 cm.sup.1) as well as DNA (1514 cm.sup.1, 1099 cm.sup.1 and 975 cm.sup.1).

(42) a) 7 Wave Number Model

(43) This model is detailed in Table 1, below. it has a calculated coefficient of determination R.sup.2 (or Multiple R-Squared) of 0.4804 and a RMSEP of 4.8%.

(44) TABLE-US-00001 TABLE 1 Wave Standard number Coefficient error Test Explanatory j Std. result Probability variable j (cm1) Estimate Error t value Pr (>|t|) 0 0.2277 0.2503 0.910 0.365812 1 2956.7 9.3246 2.4387 3.824 0.000263 2 1428.4 10.4157 4.4484 2.341 0.021764 3 1383.5 15.7746 5.2130 3.026 0.003356 4 1365.1 19.7086 5.7690 3.416 0.001011 5 1095.8 6.8730 1.4857 4.626 1.46e05 6 1036.6 7.7556 2.3304 3.328 0.001337 7 963.1 3.7342 1.1538 3.236 0.001777

(45) Distribution of residues, in minimum, maximum and quartiles:

(46) TABLE-US-00002 Minimum 1.sup.st quartile Median 3.sup.rd quartile Maximum 0.144853 0.028759 0.000373 0.033174 0.109250

(47) Residual standard error: 0.05298 with 78-degree freedom

(48) Adjusted R-Squared: 0.4337

(49) F-value or F-value of Fisher's test: 10.3 over 78-degree freedom, P-value (P-value of Fisher's test): 4.455e-09

(50) a) 9 Wave Number Model

(51) This 9 wave number model is detailed in Table 2, below. It has a calculated Multiple R-Squared of 0.5884 and a RMSEP of 4.49%.

(52) TABLE-US-00003 TABLE 2 Wave Standard number error Test Explanatory Coefficient Std. result Probability variable j (cm1) Estimate Error t value Pr (>|t|) 0 0.06038 0.30929 0.195 0.84573 1 2956.7 13.30697 2.44983 5.432 6.46e07 2 1503.9 7.95636 2.95590 2.692 0.00874 3 1444.7 7.51796 3.27999 2.292 0.02467 4 1428.4 5.34138 4.70264 1.136 0.25960 5 1383.5 11.74606 4.99481 2.352 0.02128 6 1365.1 22.37749 5.23880 4.271 5.56e05 7 1095.8 8.07636 1.48922 5.423 6.68e07 8 1036.6 9.59107 2.14903 4.463 2.76e05 9 963.1 3.14150 1.14317 2.748 0.00748

(53) Distribution of residues in minimum, maximum and quartiles:

(54) TABLE-US-00004 Minimum 1.sup.st quartile Median 3.sup.rd quartile Maximum 0.115369 0.023260 0.005633 0.021377 0.122507

(55) Residual standard error: 0.04777 with 76-degree freedom

(56) Adjusted Multiple R-Squared: 0.5397

(57) F-value: 12.07 over 76-degree freedom, P-value: 1.325e-11

(58) a) 13 Wave Number Model

(59) This model is detailed in Table 3, below. It has a Multiple R-Squared of 0.2972 and a RMSEP of 6.3%.

(60) TABLE-US-00005 TABLE 3 Wave Explana- number Coefficient Standard Test tory error result Probability variable j (cm.sup.1) Estimate Std. Error t value Pr (>|t|) 0 0.1649 0.2345 0.703 0.48275 1 3164.886 1.2917 3.4928 0.370 0.71191 2 3136.318 3.9653 4.0591 0.977 0.32979 3 3103.67 1.8260 4.33 0.422 0.67368 4 2975.115 6.4489 3.4555 1.866 0.06346 5 1736.504 3.5545 1.5065 2.359 0.01926 6 1661.004 0.244 1.339 0.182 0.85558 7 1514.085 13.7841 2.3808 5.790 2.68e08 8 1448.787 5.1977 1.584 3.281 0.00122 9 1430.422 23.5023 3.9896 5.891 1.59e08 10 1316.152 6.6227 3.62 1.829 0.06881 11 1214.124 6.0439 3.0219 2 0.04684 12 1020.273 10.1756 2.5485 3.993 9.16e05 13 975.3806 6.2081 1.4327 4.333 2.32e05

(61) Distribution of residues, in minimum. maximum and quartiles:

(62) TABLE-US-00006 Minimum 1.sup.st quartile Median 3.sup.erdquartile Maximum 0.193993 0.033218 0.001113 0.039355 0.189818

(63) Residual standard error: 0.06529 with 201-degree freedom

(64) Adjusted R-Squared: 0.2517

(65) F-value or F-value of Fisher's test): 6.537 over 13 and 201-degree of freedom, P-value (P-value of Fisher's test): 2.492e-10

(66) a) 15 Wave Number Model

(67) This model is detailed in Table 4, below. It has a Multiple R-Squared of 0.3861 and a RMSEP of 5.9%.

(68) TABLE-US-00007 TABLE 4 Wave Standard Explana- number Coefficient error Test tory Std. result Probability variable j (cm.sup.1) Estimate Error t value Pr (>|t|) 0 0.1618 0.2033 0.796 0.427213 1 3071.021 12.198 4.3248 2.82 0.005281 2 3026.129 3.2041 3.8144 0.84 0.401918 3 2977.156 6.9854 3.4328 2.035 0.043185 4 1689.572 1.1852 0.7567 1.566 0.11887 5 1618.152 4.0715 1.5196 2.679 0.007995 6 1514.085 8.8803 2.4751 3.588 0.00042 7 1430.422 21.7955 3.7797 5.767 3.05e08 8 1344.719 18.6905 6.149 3.040 0.002687 9 1338.598 38.3873 14.814 2.591 0.01027 10 1336.557 29.9668 13.9565 2.147 0.032991 11 1183.516 9.7122 3.5233 2.757 0.006384 12 1103.935 9.9713 2.9172 3.418 0.000765 13 1099.854 9.4505 2.0387 4.636 6.43e06 14 1020.273 8.8229 2.4303 3.630 0.00036 15 975.3806 2.7333 1.3692 1.996 0.047276

(69) Distribution of residues, in minimum, maximum and Quartiles:

(70) TABLE-US-00008 Minimum 1.sup.st quartile Median 3.sup.rd quartile Maximum 0.178825 0.031587 0.001598 0.036951 0.163905

(71) Residual standard error: 0.06133 with 199-degree freedom

(72) Adjusted R-Squared: 0.3398

(73) F-value or F-value of Fisher's test): 8.343 over 15 and 199-degree of freedom,

(74) P-value (P-value of Fisher's test): 1.246e-14

(75) a) 20 Wave Number Model

(76) This 20 wave number model is detailed in Table 5, below. It has a calculated Multiple R-Squared of 0.7785 and a RMSEP of 3.29%.

(77) TABLE-US-00009 TABLE 5 Wave Standard number error Test Explanatory Coefficient Std. result Probability variable j (cm1) Estimate Error t value Pr (>|t|) 0 0.2254 0.3695 0.610 0.544001 1 3091.4 24.6521 12.4717 1.977 0.052328 2 3089.4 16.4980 12.8488 1.284 0.203695 3 2987.4 10.5285 6.9744 1.510 0.135993 4 2969.0 7.9864 4.2662 1.872 0.065707 5 2956.7 9.7930 2.8089 3.486 0.000883 6 2805.7 17.1375 9.4481 1.814 0.074314 7 1520.2 4.1227 3.0247 1.363 0.177584 8 1503.9 4.8509 2.6843 1.807 0.075365 9 1452.9 0.8978 3.3514 0.268 0.789646 10 1444.7 7.9318 2.8937 2.741 0.007901 11 1428.4 15.6745 4.9713 3.153 0.002446 12 1383.5 15.7117 5.3184 2.954 0.004361 13 1365.1 12.0351 4.7273 2.546 0.013279 14 1136.6 6.2445 4.4486 1.404 0.165163 15 1124.3 0.4249 4.2978 0.099 0.921557 16 1095.8 7.5944 1.6238 4.677 1.52e05 17 1036.6 6.1430 2.5989 2.364 0.021092 18 1012.1 3.9852 3.0611 1.302 0.197555 19 963.1 2.9781 1.3443 2.215 0.030237 20 955.0 4.0846 2.1981 1.858 0.067664

(78) Distribution of residues in minimum, maximum and quartiles:

(79) TABLE-US-00010 Minimum 1.sup.st quartile Median 3.sup.rd quartile Maximum 0.068317 0.024352 0.004114 0.019129 0.086793

(80) Residual standard error: 0.03789 with 65-degree freedom

(81) Adjusted Multiple R-Squared: 0.7104

(82) F-value: 11.42 over 20 and 65-degree freedom, P-value: 2.375e-14

(83) The relevance of this 20 wave number model is illustrated in FIGS. 4A, 4B and 5.

(84) FIGS. 4A, 4B and 5 show the correlation between the NRR values predicted by the model and the values known for the samples used for the calibration (FIG. 4A), for the validation of the model (FIG. 4B) and for all the samples (FIG. 5).

(85) As can be seen in FIG. 6, for 86% of the samples, the model is accurate to less than 5 NRR points.

(86) 6.2 Exemplary Embodiment of the Invention

(87) The steps of an exemplary embodiment of a method for determining the quality of a semen of a vertebrate animal of the invention, are illustrated with reference to FIG. 1, in diagrammatic form block.

(88) At a step 110, we measure 3 times the spectrum of absorption of a prepared pellet, in a step 111, from two straws from a semen, with a Spectrometer model FT-IR SPID and with a sensor LS23 of DIAFIR (registered trademark).

(89) In a variation of this particular embodiment of the invention, the absorption spectrum can be measured by transmission, by reflection or by ATR (Attenuated Total Reflection).

(90) The 3 measured spectra are then averaged to obtain an averaged spectrum (step 112).

(91) In a step 130, a value of the second derivative of the absorption X.sub.j for n wave numbers .sub.j (1jn) characteristic of the semens is determined from the averaged absorption spectrum of the breed of the animal, selected during a step 120.

(92) Then, during a step 140, the 90-day non-return rate is calculated from the following mathematical law
Y=.sub.0+.sub.j=1.sup.n.sub.jX.sub.j
where .sub.0 and .sub.j (1jn) are constants specific to the phenotype of the a breed of the animal which produces the semen, in using the values of the second derivatives of the absorption X.sub.j.sub.(j[1;n]) determined at step 130.

(93) In variations of this particular embodiment of the invention, it may be envisaged to calculate the 90-day non-return rate from the absorption values X.sub.j for n wave numbers .sub.j (1jn) characteristic of the semens of the breed of the animal or using both values of the absorption and values of the second derivative of the absorption for the n numbers of waves .sub.j (1jn) characteristic of the semens of the breed of the animal.

(94) Although the invention has been described in connection with several particular embodiments, it is obvious that it is not limited thereto and that it comprises all the technical equivalents of the means described and their combinations if they are within the scope of the invention.

(95) Thus, the method described in relation with an example applicable to all the breeds or the species of animal vertebrates by adapting the phenotype of the breed or species through the selection of wave numbers (explanatory variables) characteristic of the quality of the semen of the breed or species.

(96) An exemplary embodiment of the invention remedies the shortcomings of the state of the art mentioned above.

(97) Specifically, an exemplary embodiment of the invention provides a method for determining the quality of a semen of a vertebrate animal that is effective and reliable any type of semen.

(98) An exemplary embodiment provides such a technique that is simple and rapid to implement.

(99) An exemplary embodiment provides method for determining the quality of a semen of a vertebrate animal, with reduced cost price.

(100) Although the present disclosure has been described with reference to one or more examples, workers skilled in the art will recognize that changes may be made in form and detail without departing from the scope of the disclosure and/or the appended claims.