Raman spectrum-based yeast seed solution culture stage rapid discrimination method

Abstract

The invention discloses a method for rapidly discriminating the culture stage of a yeast seed solution based on a Raman spectrum. The method specifically comprises the following steps: selecting a yeast single colony and culturing in a YPD test tube to prepare a primary seed solution; inoculating the primary seed solution into a shake flask containing 200 mL of a YPD liquid culture medium for shake culture; in the shake-flask culture process, a 785 nm Raman spectrometer is used for in-situ collection of a Raman spectrum, and sampling is carried out every 30 min to determine OD600; and establishing a yeast seed solution culture stage detection model based on the collected spectral data and OD600 data. According to the method, frequent sampling is not needed, the method can be used for rapidly distinguishing the hysteresis phase, the logarithmic growth phase and the stable phase of the yeast seed liquid, the optimal inoculation opportunity can be scientifically determined, and the consistency of the fermentation process and the product quality stability are improved.

Description

[0001] The present invention relates to seed liquid culture detection technology in the field of biological fermentation, and in particular to a method for quickly distinguishing the culture stage of yeast seed liquid based on Raman spectroscopy. The method can be used to quickly distinguish whether the yeast seed liquid is in the lag phase, logarithmic growth phase or stable phase, thereby determining the optimal inoculation time and improving the consistency of main fermentation batches and the stability of product quality.

[0002] In industrial fermentation production, a multi-stage scale-up inoculation model is often employed to ensure efficient start-up of the primary fermentation and the accumulation rate of the target product. This model typically begins with a single colony purified from a slant or plate, progressing through primary seed solution and shake flask culture, to small seed tanks, medium seed tanks, and finally to the primary fermentation tank. This staged scale-up effectively shortens the primary fermentation cycle, reduces uncontrollable factors, and improves the reproducibility and stability of the overall production process.

[0003] Among them, the growth state of the seed liquid in shake flask culture directly determines the initial metabolic path after inoculation of the fermenter and the consistency of the overall production. If the inoculation time is too early or too late, it may lead to an extension of the fermentation cycle or metabolic abnormalities, thereby affecting product quality. Usually, the best time to inoculate the seed liquid is in the late logarithmic growth phase or the early stationary phase. At this time, the cell concentration is high and the metabolic activity is strong, which is most suitable for fermentation tank scale-up inoculation. Therefore, in actual production, scientifically determining the culture stage of the seed liquid will help to scientifically determine the best time to inoculate, thereby effectively reducing the abnormal fermentation batch rate and improving the overall product yield.

[0004] Currently, methods used to determine the culture stage of a seed solution during fermentation mainly include the OD600 method, microscopic observation, and biochemical indices. The OD600 turbidimetric method uses an ultraviolet spectrophotometer to measure the absorbance of cells at a wavelength of 600 nm to reflect cell density. By plotting growth curves based on multiple time-point measurements, cells can be distinguished as being in the lag phase, logarithmic growth phase, or stationary phase. This method is simple to operate, but it only reflects cell concentration and suffers from significant errors during the logarithmic growth phase. The microscopic observation method observes cell morphology and division status under a microscope after sampling, and then determines the growth stage based on the experimenter's experience. This method is highly dependent on the operator's subjective experience, has poor reproducibility, and is difficult to implement for rapid, large-scale testing. The biochemical indices method indirectly reflects cellular metabolic activity by measuring changes in the concentration of biochemical indicators such as substrates and metabolites in the seed solution, thereby inferring the culture stage. However, this method typically requires offline testing, resulting in long detection cycles and difficulty in timely determining the optimal timing for seed solution inoculation.

[0005] Raman spectroscopy is an analytical detection technique based on molecular vibration scattering, which has the advantages of no sample pretreatment, in-situ online acquisition, weak water interference and simultaneous multi-parameter information detection compared with infrared spectroscopy and ultraviolet spectroscopy. In recent years, with the continuous progress of laser, spectrometer and detector technologies, Raman spectroscopy has gradually increased in research and application in many fields such as biology, food, medicine and environment. Especially in the field of biological fermentation, Raman spectroscopy can be used to monitor the dynamic changes of substrates, products and metabolic intermediates in the fermentation broth online, and can be combined with chemometrics methods to construct multi-parameter prediction models to realize the comprehensive analysis of the fermentation process.

[0006] The technical problem to be solved by the present application is to provide a method for rapidly distinguishing the culture stage of yeast seed liquid based on Raman spectroscopy, which can rapidly distinguish the lag phase, logarithmic growth phase and stationary phase of the yeast seed liquid without frequent sampling, and help to scientifically determine the best inoculation time, thereby improving the batch consistency and product quality stability.

[0007] To solve the above technical problems, the present application adopts the following technical solutions:

[0008] A method for rapidly distinguishing the culture stage of yeast seed liquid based on Raman spectroscopy, comprising the following steps.

[0009] (1) picking a single colony of yeast from a plate, inoculating 5 mL of liquid YPD test tube, and culturing overnight at 30℃, 150 rpm to prepare a primary seed liquid;

[0010] (2) inoculating the primary seed liquid into a flask containing 200 mL of YPD medium, with an inoculation amount of 3%, and culturing at 30℃, 150 rpm;

[0011] (3) during the flask culture, using a 785 nm Raman optical fiber probe to collect Raman spectra in-situ, and using an ultraviolet spectrophotometer to measure OD at every 30 min to obtain the measured value;

[0012] (4) sequentially performing median filtering, adaptive iteratively reweighted penalized least squares (airPLS) baseline correction, Savitzky-Golay (SG) smoothing processing and characteristic peak processing on the collected Raman spectra;

[0013] (5) based on the spectral data and the corresponding OD600 measured value, using PLS1 algorithm to establish a detection model for distinguishing the lag phase, logarithmic growth phase and stationary phase; and using the model for real-time Raman spectrum prediction of subsequent batches, and combining the model prediction results to distinguish the culture stage of the yeast seed liquid by the experimenters to scientifically determine the best inoculation time.

[0014] Furthermore, in the step (1), the single colony of Saccharomyces cerevisiae used is derived from a morphologically single colony obtained by multiple dilution and streak purification on a YPD plate. The formula of the YPD medium is 20 g / L yeast extract, 10 g / L peptone, and 20 g / L glucose. All culture media are sterilized at 121° C. and then naturally cooled for use to prevent contamination by other bacteria.

[0015] Furthermore, in step (3), the Raman fiber optic probe used is an immersion probe that can be directly inserted into the seed liquid to continuously acquire the Raman spectrum in situ; the Raman spectrometer used has an excitation wavelength of 785 nm, a laser power of 30 mW, an integration time of 15 s, and an average scan of 10 times to ensure that the spectral signal is stable and reliable.

[0016] Furthermore, in step (4), the spectrum preprocessing method includes the following four steps: first, the Raman original spectrum is processed by median filtering with a window width of 3 points to eliminate the spike signal caused by occasional cosmic rays; then, the baseline is corrected using the adaptive iterative reweighted penalized least squares method (airPLS), and background drift removal is achieved by constructing a second-order difference matrix and introducing an asymmetric weighting function, combined with an iterative update strategy; then, the spectrum is smoothed using the Savitzky-Golay (SG) method with a window of 15 and a fitting second-order polynomial; finally, the characteristic peak at the sapphire window of the Raman probe is selected as the internal standard, and each spectrum is normalized to eliminate the overall light intensity deviation between samples.

[0017] Furthermore, in step (5), the PLS1 algorithm is used to construct a quantitative prediction model between Raman spectroscopy and OD600. During the model training process, the optimal number of latent variables is determined by 5-fold cross validation, and the coefficient of determination R is used to determine the optimal number of latent variables. 2 The model performance is comprehensively evaluated by multiple indicators such as modeling error RMSEC and cross-validation error RMSECV to ensure the fitting accuracy and generalization ability of the model.

[0018] Compared with the prior art, the present invention has the following beneficial effects:

[0019] The method described in the present invention achieves non-destructive identification of the seed liquid culture stage through online spectral data collection and modeling. Compared with traditional methods that require frequent offline sampling, the present invention can obtain information on the growth status of the seed liquid without interrupting the culture process, improving detection efficiency and reducing the risk of contamination. The established discrimination model can provide a scientific basis for experimental operations, effectively reducing the inconsistencies caused by manual experience, ensuring accurate identification of the optimal inoculation time, thereby reducing the rate of abnormal fermentation batches and improving the stability of the main fermentation stage and the consistency of product quality.

[0020] In order to more clearly illustrate the technical solution of the present invention, the present invention will be further described below with reference to the accompanying drawings, but the present invention is not limited to the specific forms shown in the drawings.

[0021] Figure 1 This is a schematic diagram of the structure of the device for Raman spectroscopy detection of seed liquid. The numbers in the figure are: 1 shake flask; 2 shaker; 3 Raman fiber optic probe; 4 Raman analyzer; 5 notebook.

[0022] Figure 2 The figure is a schematic diagram of the operation flow of the method of the present invention.

[0023] Figure 3 This is the OD600 change curve of yeast seed liquid at different time points during shake flask culture.

[0024] Figure 4 The graph of model prediction results and measured values ​​changing over time

[0025] The method for rapidly distinguishing the culture stage of yeast seed liquid based on Raman spectroscopy according to the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings, but the present invention is not limited to the following embodiments.

[0026] like Figure 1 As shown, the detection system mainly includes a shake flask 1, a shaker 2, a Raman fiber optic probe 3, a Raman analyzer 4 and a notebook 5, wherein the shake flask 1 is placed on the shaker 2 for oscillation culture, the Raman fiber optic probe 3 is connected to the Raman analyzer 4 through an optical fiber, and the Raman analyzer 4 is further connected to the notebook 5 to realize real-time acquisition and analysis of spectral signals.

[0027] like Figure 2 As shown, the operational process of the method described in the present invention includes the following steps: First, the selected Saccharomyces cerevisiae strain is purified by repeated dilution and streaking on a YPD plate before use. The YPD medium used is formulated as 20 g / L yeast extract, 10 g / L peptone, and 20 g / L glucose. All media are autoclaved at 121°C for 15 minutes and then placed in a clean bench to cool naturally before use. Subsequently, within the clean bench, a single yeast colony with intact morphology and no contamination is picked from the plate using an inoculating loop that has been thoroughly sterilized and cooled with an alcohol burner flame. The colony is then inoculated into a 15 mL test tube prefilled with sterile liquid YPD medium. Gently rotate the tube to fully suspend the colony in the medium. The test tube is then placed in a constant temperature shaking incubator at 30°C and 150 rpm for overnight incubation to produce a primary liquid seed solution.

[0028] After the completion of the primary seed liquid culture, the primary seed liquid was taken with a sterile pipette in a clean bench, and was quickly inoculated into a 250 mL conical flask 1 containing 200 mL of sterilized YPD liquid medium. The flask 1 was placed on a shaker 2. Then, the inoculated flask 1 was placed in the shaker 2 set at 150 rpm for flask culture. In the process of flask culture, the immersion Raman probe was directly inserted into the seed liquid for in-situ Raman spectrum collection. Specifically, the Raman fiber probe 3 was inserted into the seed liquid in the flask 1, and the Raman fiber probe 3 was connected with a Raman analyzer 4, and the Raman analyzer 4 was connected to a notebook computer 5 through a USB interface for data collection and control. The Raman spectrum collection parameters were set as follows: laser power 30 mW, integration time 15 s, and average scanning 10 times. In the process of culture, a group of Raman spectra was automatically collected every 1.5 min, and the data file was stored. At the same time, in the above-mentioned flask culture process, 2 mL of seed liquid sample was taken from the flask 1 every 30 min using a sterile pipette, and the absorbance was immediately measured at 600 nm wavelength using a UV spectrophotometer. Before measurement, the blank control was zeroed with sterile YPD liquid medium. If the expected sample absorbance exceeds 1.0, it is necessary to dilute it with 5 or 10 times of sterile YPD liquid medium, so that the OD600 reading is within the linear detection range of 0.1-1.0. Each sample was detected in parallel for 3 times, and the average value was calculated. As shown in FIG. 1, the OD600 value measured in the process of flask culture showed a typical curve of first flat, then rapid rise and then stable with the increase of time, which could be divided into lag phase, logarithmic growth phase and stable phase. Figure 3

[0029] In the process of 12-hour flask culture, 2880 Raman spectrum data were collected in-situ at an interval of 15 seconds; at the same time, OD600 was measured offline every 30 minutes to obtain 24 measured points, and 2880 corresponding OD600 values consistent with the Raman collection time points were calculated by interpolation method. Finally, 2880 groups of sample data were formed, each group containing one pretreated spectrum and one corresponding OD600 value.

[0030] ​The raw Raman spectral data were first median filtered to remove occasional spike noise. Then, baseline correction was performed using the adaptive iteratively reweighted penalized least squares (airPLS) method to eliminate fluorescence drift. The resulting spectral data matrix was then mean-centered and standard-deviation normalized to eliminate the effects of wavelength-varying scales and improve the stability and consistency of PLS1 model training. The collected raw Raman spectral data were preprocessed in the following steps: first, median filtering was used to remove occasional spike noise, and the filter window width was set to 3; then, the adaptive iterative reweighted penalized least squares (airPLS) algorithm was used for baseline correction, and the background curve was iteratively estimated by constructing a second-order difference matrix and a residual weighting function to eliminate fluorescence interference; then, the Savitzky-Golay (SG) method was used for spectral smoothing with a window of 15 and a fitting polynomial order of 2 to preserve the main peak morphology while suppressing high-frequency noise; finally, all spectra were normalized using the stable characteristic peak intensity at the sapphire window of the Raman probe as the internal standard to eliminate intensity deviations caused by sampling errors or light source drift, thereby ensuring the consistency and stability of model training.

[0031] The multidimensional spectral matrix after preprocessing was used as the independent variable X, and the OD600 value was used as the response variable Y. A 5-fold cross-validation was used to determine the optimal number of latent variables and establish a PLS1 prediction model. The determination coefficient R 2 The RMSEC is 0.96, the RMSEC is 0.05, and the RMSECV is 0.07, indicating that the model has high fitting accuracy and stability, and can predict the OD600 value in real time in subsequent culture batches, providing an objective basis for experimenters to distinguish the yeast growth stage. Subsequently, the Raman spectral data collected in real time in subsequent culture batches were input into the PLS1 model for prediction. Figure 4 As shown in the figure, the model's predictions are highly consistent with the actual OD600 value trends throughout the culture process. Using the model's output, researchers can quickly determine whether the seed culture stage is the lag phase, logarithmic growth phase, or stationary phase, thereby scientifically determining the optimal inoculation time.

Claims

1. A method for rapidly identifying the culture stage of yeast seed liquid based on Raman spectroscopy, characterized in that: The following steps are involved: (1) Pick a single yeast colony from the plate, inoculate it into a 5 mL liquid YPD test tube, and culture it at 30°C and 150 rpm overnight to prepare the first-level seed solution; (2) The primary seed solution was inoculated into a shake flask containing 200 mL of YPD medium with an inoculation volume of 3%, and cultured at 30°C and 150 rpm; (3) During the shake flask culture process, a 785 nm Raman fiber optic probe was used to collect Raman spectra in situ, and an ultraviolet spectrophotometer was used to measure the OD values ​​every 30 min. (4) The collected Raman spectra were subjected to median filtering, adaptive iterative reweighted penalized least squares (airPLS) baseline correction, Savitsky-Golay (SG) smoothing and characteristic peak processing in sequence; (5) Based on the spectral data and the corresponding OD600 measured values, the PLS1 algorithm was used to establish a detection model for distinguishing the lag phase, logarithmic growth phase, and stationary phase; and the model was used for real-time Raman spectral prediction of subsequent batches. The experimenters combined the model prediction results to determine the culture stage of the yeast seed liquid in order to scientifically determine the optimal inoculation time.

2. The method according to claim 1, characterized in that The single colony used in step (1) is derived from a single morphological colony obtained by purifying yeast multiple times by streak dilution method on a YPD plate.

3. The method according to claim 1, characterized in that The Raman fiber optic probe used in step (3) is an immersion Raman probe that can be directly inserted into the seed liquid to continuously acquire the spectrum in situ.

4. The method according to claim 1, wherein The excitation wavelength of the Raman spectrometer used in step (3) is 785 nm, the laser power is 30 mW, the integration time is 1.5 s, and the average scan is 10 times.

5. The method according to claim 1, wherein The test wavelength of the UV spectrophotometer used in step (3) is 600 nm. Sterile YPD culture medium is used to adjust the sample to zero before measurement. Each sample is tested in parallel 3 times and the average value is taken as the actual OD600 value.

6. The method according to claim 1, characterized in that The spectral preprocessing method used in step (4) includes: using a median filter with a window width of 3 points to remove occasional spike noise; using an adaptive iterative reweighted penalized least squares method (airPLS) for baseline correction, and iteratively optimizing by constructing a second-order difference matrix and a residual weighting function to remove background drift; then using the Savitzky-Golay (SG) method with a window width of 15 points and a fitting order of 2 to smooth the spectrum, and using the intensity of the stable characteristic peak at the sapphire window as an internal standard to normalize all spectra.

7. The method according to claim 1, characterized in that In step (5), the experimenter manually determines the culture stage of subsequent batches of yeast seed liquid based on the prediction results of the PLS1 model to scientifically determine the optimal inoculation time.

Images