A method for synthesizing penicillin V potassium via enzyme catalysis
By adjusting the catalytic control parameters in multiple dimensions, the problem of low penicillin V potassium generation efficiency in enzyme catalysis was solved, and a more efficient catalytic process was achieved.
Patent Information
- Application Number
- CN202611240235.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-17
- Publication Date
- 2026-11-17
AI Technical Summary
In the existing enzyme-catalyzed synthesis of penicillin V potassium, the catalytic control reference dimension is singular and lacks a multi-stage control scheme, resulting in low production efficiency.
By acquiring various raw parameter data and liquid phase detection signals, impurity information is identified, the catalytic stage is determined, and the catalytic control parameter vector is adjusted according to key parameters to achieve multi-dimensional control.
It improved the production efficiency and quality of penicillin V potassium and enhanced the accuracy of catalytic control.
Smart Images

Figure CN123108101A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of enzyme catalysis technology, and in particular to a method for synthesizing potassium penicillin V by enzyme catalysis. Background Technology
[0002] Penicillin V potassium (fluorobenzylpenicillin potassium) is an important β-lactam antibiotic, widely used to treat Gram-positive bacterial infections. Its traditional synthesis mainly adopts an enzymatic catalytic method: using 6-aminopenicillanic acid (6-APA) and methyl phenoxyacetate as substrates, the enzyme catalyzes the reaction under the catalysis of penicillin V acylase (PVA) to produce penicillin V potassium.
[0003] However, in the process of synthesizing penicillin V potassium using existing enzyme catalysis methods, catalysis control is achieved only by periodically collecting a single parameter (such as pH or temperature). Therefore, the reference dimension for catalysis control is too singular, and there is a lack of catalysis control schemes for different catalytic stages, resulting in insufficient catalysis control of the enzyme catalysis process, which ultimately seriously affects the production efficiency of penicillin V potassium. Summary of the Invention
[0004] To solve the above-mentioned technical problems, this invention provides a method for synthesizing penicillin V potassium via enzyme catalysis. The technical solution of this invention is as follows: A method for synthesizing penicillin V potassium via enzyme catalysis, comprising: S1, acquire multiple raw parameter data and liquid phase detection signals at each sampling moment from the catalytic start point to the current sampling moment during the penicillin V potassium enzyme catalysis process; S2, based on multiple raw parameter data and liquid phase detection signals at each sampling time, identifies multiple impurity information at each sampling time during the penicillin V potassium enzyme catalysis process; S3, determine the catalytic stage at the current sampling time based on various impurity information at each sampling time; S4. Based on the catalytic stage at the current sampling time and the various raw parameter data at each sampling time, select various key parameters from all raw parameters and obtain the various key parameter data for each sampling time for each key parameter. S5, based on the multiple key parameter data at each sampling time and the catalytic stage at the current sampling time, adjust the catalytic control parameter vector for the next sampling time until the penicillin V potassium enzyme catalysis process ends.
[0005] Preferably, S2 includes: S21, convert the liquid phase detection signal at each sampling time into a chromatogram at each sampling time, and obtain multiple impurity chromatographic peaks in the chromatogram at each sampling time; S22, compare multiple impurity chromatographic peaks in the chromatogram at each sampling time with a pre-set impurity database to obtain the impurity corresponding to each impurity chromatographic peak at each sampling time; S23, calculate the peak area of the impurity chromatographic peak corresponding to each impurity at each sampling time, and calculate the first content of each impurity at each sampling time based on the peak area of the impurity chromatographic peak corresponding to each impurity at each sampling time; S24, determine the types of multiple impurities at each sampling time, and construct the metabolic network for each sampling time based on the types of multiple impurities at each sampling time and the multiple original parameter data at each sampling time, and calculate the second content of each impurity at each sampling time based on the metabolic network at each sampling time; S25, calculate the content of each impurity at each sampling time based on the first and second contents of each impurity at each sampling time, and construct multiple impurity information at each sampling time from the type and content of each impurity at each sampling time.
[0006] Preferably, S24 includes: S241, standardize the multiple raw parameter data at each sampling time to obtain multiple standard parameter data at each sampling time; S242, calculates the oxygen uptake rate, carbon dioxide release rate, respiratory quotient and dynamic derivative at each sampling time based on multiple standard parameter data at each sampling time, and determines the main cause of impurity generation at each sampling time based on the oxygen uptake rate, carbon dioxide release rate, respiratory quotient and dynamic derivative at each sampling time. S243, determine the types of multiple impurities at each sampling time based on the main cause of impurity generation at each sampling time, and generate the instantaneous content of each impurity at each sampling time based on the types of multiple impurities at each sampling time and the multiple standard parameter data at each sampling time; S244, construct the metabolic network for each sampling time based on the instantaneous content of each impurity at each sampling time; S245, calculate the cumulative content of each impurity at each sampling time based on the metabolic network at each sampling time, and use the cumulative content of each impurity as the second content of each impurity at each sampling time.
[0007] Preferably, the main causes of impurity generation include impurities generated by metabolic byproducts, impurities generated by oxidation, or impurities generated under normal conditions, and S242 includes: S2421, calculates the oxygen uptake rate at each sampling time based on the intake flow rate data, catalytic liquid volume data, intake oxygen concentration data and exhaust oxygen concentration data from multiple standard parameter data at each sampling time. S2422, calculates the carbon dioxide release rate at each sampling time based on the intake flow rate data, catalyst volume data, intake carbon dioxide concentration data and exhaust carbon dioxide concentration data from multiple standard parameter data at each sampling time. S2423, calculate the ratio of carbon dioxide release rate to oxygen uptake rate at each sampling time to obtain the respiratory quotient at each sampling time; S2424, calculate the derivative of the respiratory quotient with respect to time at each sampling moment to obtain the dynamic derivative at each sampling moment; S2425, when the oxygen uptake rate at any sampling time is greater than a preset oxygen uptake rate threshold, the carbon dioxide release rate remains unchanged compared to the previous sampling time, the respiratory quotient decreases compared to the previous sampling time, and the dynamic derivative is negative, the main cause of impurity generation at any sampling time is determined to be impurities generated by oxidation; when the oxygen uptake rate remains unchanged compared to the previous sampling time, the carbon dioxide release rate increases compared to the previous sampling time, the respiratory quotient increases compared to the previous sampling time, and the dynamic derivative is positive, the main cause of impurity generation at any sampling time is determined to be impurities generated by metabolic byproducts; otherwise, the impurity generation at any sampling time is determined to be a normal state.
[0008] Preferably, S3 includes: S31, calculate the proportion of each impurity at each sampling time based on the multiple impurity information at each sampling time; S32, construct a proportion time series based on the proportion of each impurity at each sampling time; S33 determines multiple turning points based on the proportion time series, and determines the catalytic stage at the current sampling time based on the multiple turning points.
[0009] Preferably, S33 includes: S331, map the percentage time series to a two-dimensional phase plane, where each element in the percentage time series is mapped to a point in the two-dimensional phase plane; S332 connects points representing the same impurity at adjacent sampling times on a two-dimensional phase plane to form multiple continuous trajectory curves; S333, for each point on each continuous trajectory curve, calculate its first-order difference and second-order difference, and calculate the curvature of each point on each continuous trajectory curve based on the first-order difference and second-order difference of each point. S334, based on the curvature of each point on each continuous trajectory curve, obtain the points corresponding to the local curvature maxima on multiple continuous trajectory curves as multiple inflection points, and filter multiple valid inflection points from multiple inflection points according to the curvature threshold; S335 divides multiple effective inflection points into multiple catalytic stages according to their curvature and determines the catalytic stage at the current sampling moment.
[0010] Preferably, S4 includes: S41, construct the main reaction pathway at the current sampling time based on the catalytic stage at the current sampling time, wherein the main reaction pathway includes multiple directed edges and their flux weights; S42, select multiple main edges from all directed edges based on the flux weight of each directed edge and the preset activity threshold; S43, For each original parameter, calculate the mutual information between each original parameter and each main edge based on the original parameter data of each original parameter at each sampling time, and select multiple original parameters whose mutual information is greater than the preset mutual information threshold as multiple key parameters; S44, selects each key parameter from the multiple raw parameter data at each sampling time.
[0011] Preferably, S41 includes: S411, construct a set of metabolites based on the impurity types in the various impurity information at the current sampling time, and construct a set of reaction nodes based on the catalytic stage at the current sampling time; S412 connects each reaction node in the set of reaction nodes to each metabolic node in the set of metabolites, forming each directed edge; S413, calculate the flux weight of each directed edge based on the impurity content in the various impurity information at the current sampling time and the catalytic stage at the current sampling time; S414 is the main reaction path at the current sampling time, which is composed of each directed edge and its flux weight.
[0012] Preferably, when S5 adjusts the catalytic control parameter vector for the next sampling time based on multiple key parameter data at each sampling time and the catalytic stage at the current sampling time, it includes: S51, standardize the multiple key parameter data of each sampling moment to obtain multiple standard key parameter data of each sampling moment, and obtain the starting sampling moment of the catalytic stage at the current sampling moment. S52, construct a state transition function based on the preset response coefficient matrix corresponding to the catalytic stage at the current sampling time and the various standard key parameter data of each sampling time from the stage start sampling time to the current sampling time; S53, Based on the various standard key parameter data and state transition function at the current sampling time, determine the predicted key parameter data for the next sampling time; S54, calculate the deviation vector between the standard key parameter data and the predicted key parameter data at the current sampling time; S55, determine the catalytic control parameter adjustment vector for the next sampling time based on the deviation vector and the pre-trained PID model; S56. The catalytic control parameter adjustment vector at the next sampling time is superimposed on the catalytic control parameter vector at the current sampling time to obtain the catalytic control parameter vector at the next sampling time.
[0013] All of the above-mentioned optional technical solutions can be combined arbitrarily, and the present invention will not provide a detailed description of the structure after each combination.
[0014] By means of the above solution, the beneficial effects of the present invention are as follows: This invention identifies multiple impurity information at each sampling moment during the penicillin V potassium enzyme catalysis process. Based on this, it determines the catalytic stage at the current sampling moment and then filters multiple key parameter data from the various raw parameter data at each sampling moment. Combined with the current catalytic stage, it adjusts the catalytic control parameter vector for the next sampling moment in the penicillin V potassium enzyme catalysis process. Compared with traditional methods, this invention not only determines the stage of the penicillin V potassium enzyme catalysis process through impurity information, but also filters multiple key parameters affecting the penicillin V potassium enzyme catalysis process through the catalytic stage. This increases the dimension of catalytic control reference, thereby improving the accuracy of subsequent catalytic control adjustments and improving the production efficiency and quality of penicillin V potassium.
[0015] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, the preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings. Attached Figure Description
[0016] Figure 1 This is a schematic flowchart of a method for synthesizing penicillin V potassium by enzyme catalysis provided by the present invention. Detailed Implementation
[0017] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.
[0018] like Figure 1 As shown in the embodiment of the present invention, a method for synthesizing penicillin V potassium by enzyme catalysis includes: S1, acquire multiple raw parameter data and liquid phase detection signals at each sampling moment from the catalytic start point to the current sampling moment during the penicillin V potassium enzyme catalysis process.
[0019] Specifically, the types of raw parameters included in the data collection included inlet air flow rate, catalytic liquid volume, inlet oxygen concentration, exhaust oxygen concentration, inlet carbon dioxide concentration, exhaust carbon dioxide concentration, and pH value. Liquid phase detection signals were acquired using a liquid phase analyzer. Multiple raw parameter data at each sampling time were synchronously acquired using acquisition devices with the same sampling frequency. Specifically, raw inlet air flow rate data was acquired using a flow sensor, while raw inlet oxygen concentration data and raw exhaust oxygen concentration data were acquired using an oxygen detection probe. The liquid phase analyzer, flow sensor, and oxygen detection probe all had the same acquisition frequency and were synchronized in time.
[0020] S2 identifies multiple impurity information at each sampling moment based on various raw parameter data and liquid phase detection signals during the penicillin V potassium enzyme catalysis process.
[0021] Specifically, the time interval between adjacent sampling times in this embodiment of the invention is 5-10 seconds. Various impurity information includes impurity type and impurity content. Impurity types include: phenoxymethylpenicillin thiazolic acid, phenoxyacetic acid, 4-hydroxypenicillin V potassium, and penicillin G potassium, etc.
[0022] In one specific embodiment, S2 includes: S21, convert the liquid phase detection signal at each sampling time into a chromatogram at each sampling time, and obtain multiple impurity chromatographic peaks in the chromatogram at each sampling time.
[0023] Specifically, the liquid chromatography detection signal at a given sampling time refers to the liquid chromatography detection signal collected within the time interval between that sampling time and the previous sampling time. A chromatographic conversion system transforms the liquid chromatography detection signal at each sampling time into a chromatogram for that sampling time. In the chromatogram's coordinate system, the horizontal axis represents time (typically in 1 milliseconds), and the vertical axis represents signal intensity. Then, an automatic peak detection algorithm identifies multiple impurity chromatographic peaks in the chromatogram at each sampling time.
[0024] S22, compare the multiple impurity chromatographic peaks in the chromatogram at each sampling time with a pre-set impurity database to obtain the impurity corresponding to each impurity chromatographic peak at each sampling time.
[0025] Specifically, firstly, the time interval corresponding to each impurity chromatographic peak in the chromatogram at each sampling time is obtained. The method for obtaining the time interval corresponding to a specific impurity chromatographic peak in the chromatogram at a given sampling time is as follows: the point where the signal intensity to the left of the impurity chromatographic peak decreases to half the peak value is taken as the left endpoint, and the point where the signal intensity to the right of the impurity chromatographic peak decreases to half the peak value is taken as the right endpoint. The left and right endpoints constitute the time interval corresponding to that impurity chromatographic peak at that sampling time. Then, based on the length of the time interval corresponding to each impurity chromatographic peak and its peak value, the characteristics of each impurity chromatographic peak are used as features. The characteristics of each impurity chromatographic peak are compared with the standard features (length of the standard time interval and standard peak value) of each impurity in a pre-defined impurity database to obtain the impurity corresponding to each impurity chromatographic peak. Specifically, when comparing the characteristics of a certain impurity chromatographic peak with the standard characteristics of a certain impurity, the absolute value of the difference between the length of the time interval of the characteristic of the impurity chromatographic peak and the length of the standard time interval of the standard characteristics of the impurity is calculated to obtain the time interval difference. The absolute value of the difference between the peak value of the characteristic of the impurity chromatographic peak and the standard peak value of the standard characteristics of the impurity is calculated to obtain the peak value difference. When the time interval difference is less than 2% of the time interval length corresponding to each impurity chromatographic peak, and the peak value difference is less than 2% of the peak value of each impurity chromatographic peak, and both the time interval difference and the peak value difference are the minimum values of the impurity chromatographic peak and the impurity of that type among all types of impurities, the impurity corresponding to the impurity chromatographic peak is determined to be that type of impurity.
[0026] S23, calculate the peak area of the impurity chromatographic peak corresponding to each impurity at each sampling time, and calculate the first content of each impurity at each sampling time based on the peak area of the impurity chromatographic peak corresponding to each impurity at each sampling time.
[0027] Specifically, the area enclosed by the impurity chromatographic peak corresponding to each impurity at each sampling time and the x-axis (time axis) is calculated to obtain the peak area of the impurity chromatographic peak corresponding to each impurity. When calculating the first content of a certain impurity at a certain sampling time based on the peak area of the impurity chromatographic peak corresponding to a certain impurity at a certain sampling time, the peak area of the penicillin V potassium chromatographic peak at that sampling time is first obtained (the principle of obtaining the peak area is the same as the calculation principle of the peak area of the impurity chromatographic peak mentioned above). The ratio of the peak area of the penicillin V potassium chromatographic peak to the peak area of the impurity chromatographic peak corresponding to that impurity is calculated to obtain the comparison factor. Then, the comparison factor is multiplied by the preset response factor of that impurity to obtain the first content of that impurity at that sampling time.
[0028] S24, determine the types of multiple impurities at each sampling time, and construct the metabolic network for each sampling time based on the types of multiple impurities at each sampling time and the multiple original parameter data at each sampling time. Calculate the second content of each impurity at each sampling time based on the metabolic network at each sampling time.
[0029] Specifically, in this embodiment of the invention, the metabolic network is a network describing the metabolic impurities of the penicillin V potassium catalytic process.
[0030] In one specific embodiment, S24 includes: S241 standardizes the various raw parameter data at each sampling time to obtain various standard parameter data at each sampling time.
[0031] Specifically, when standardizing the multiple raw parameter data at each sampling time, a unified data standardization operation is performed on the multiple raw parameter data from the catalytic start point to the current sampling time (standardized to the [0,1] interval).
[0032] S242 calculates the oxygen uptake rate, carbon dioxide release rate, respiratory quotient, and dynamic derivative at each sampling time based on multiple standard parameter data at each sampling time, and determines the main cause of impurity generation at each sampling time based on the oxygen uptake rate, carbon dioxide release rate, respiratory quotient, and dynamic derivative at each sampling time.
[0033] The main causes of impurities include impurities generated by metabolic byproducts, impurities generated by oxidation, or impurities generated in a normal state.
[0034] In one specific embodiment, S242 includes: S2421 calculates the oxygen uptake rate at each sampling time based on the intake flow rate, catalytic liquid volume, intake oxygen concentration, and exhaust oxygen concentration data from various standard parameter data at each sampling time.
[0035] Specifically, in calculating the oxygen uptake rate at a given sampling time... When, the formula is: ;in, The molar coefficient represents the conversion of gas volume into the number of moles of gas, and V represents the volume of the catalytic liquid at that sampling time. This represents the intake airflow data at that sampling moment. This represents the intake oxygen concentration data at that sampling time. This indicates the oxygen concentration data in the exhaust gas at the sampling time.
[0036] S2422 calculates the carbon dioxide release rate at each sampling time based on the intake flow rate, catalytic liquid volume, intake carbon dioxide concentration, and exhaust carbon dioxide concentration data from various standard parameter data at each sampling time.
[0037] Specifically, based on the above, the carbon dioxide release rate at this sampling time is calculated. When, the formula is: ;in, This represents the carbon dioxide concentration data in the exhaust gas at that sampling time. This indicates the intake carbon dioxide concentration data at that sampling moment.
[0038] S2423, calculate the ratio of carbon dioxide release rate to oxygen uptake rate at each sampling time to obtain the respiratory quotient at each sampling time.
[0039] S2424, calculate the time derivative of the respiratory quotient at each sampling time to obtain the dynamic derivative at each sampling time.
[0040] Specifically, in this embodiment of the invention, the dynamic derivative at each sampling moment is calculated by calculating the difference between the relative times of two adjacent sampling moments. Therefore, the formula for calculating the dynamic derivative E at a certain sampling moment is: Where RQ represents the respiratory quotient at that sampling moment, This represents the respiratory quotient of the previous sampling time. This represents the time interval between two adjacent sampling times.
[0041] S2425, when the oxygen uptake rate at any sampling time is greater than a preset oxygen uptake rate threshold, the carbon dioxide release rate remains unchanged compared to the previous sampling time, the respiratory quotient decreases compared to the previous sampling time, and the dynamic derivative is negative, the main cause of impurity generation at any sampling time is determined to be impurities generated by oxidation; when the oxygen uptake rate remains unchanged compared to the previous sampling time, the carbon dioxide release rate increases compared to the previous sampling time, the respiratory quotient increases compared to the previous sampling time, and the dynamic derivative is positive, the main cause of impurity generation at any sampling time is determined to be impurities generated by metabolic byproducts; otherwise, the impurity generation at any sampling time is determined to be a normal state.
[0042] S243, determine the types of multiple impurities at each sampling time based on the main cause of impurity generation at each sampling time, and generate the instantaneous content of each impurity at each sampling time based on the types of multiple impurities at each sampling time and the multiple standard parameter data at each sampling time.
[0043] Specifically, the types of impurities produced by metabolic byproducts include: phenoxymethylpenicillin thiazolic acid, penicillin G potassium, and phenoxyacetic acid; the types of impurities produced by oxidation include: 4-hydroxypenicillin V potassium, penicillin G potassium, and phenoxyacetic acid; and the types of impurities in the normal state include: penicillin G potassium and phenoxyacetic acid.
[0044] An impurity metabolism model is trained based on a large amount of historical standard parameter data and the instantaneous content of various historical impurities. The standard parameter data and the types of various impurities at each sampling time are input into the trained impurity metabolism model, which then generates the instantaneous content of each impurity at each sampling time. The impurity metabolism model can be a deep learning model. In this embodiment of the invention, environmental parameters (e.g., temperature and pH) during the production and training processes are set to standard environmental parameters.
[0045] S244, construct the metabolic network for each sampling time based on the instantaneous content of each impurity at each sampling time.
[0046] Specifically, the metabolic network at a certain sampling time includes multiple nodes. Each node represents the type and instantaneous content of each impurity at each sampling time from the catalytic initiation point to that sampling time. Among all nodes at that sampling time, multiple nodes belonging to the same impurity are connected to form the edges for each impurity.
[0047] S245, calculate the cumulative content of each impurity at each sampling time based on the metabolic network at each sampling time, and use the cumulative content of each impurity as the second content of each impurity at each sampling time.
[0048] Specifically, when calculating the cumulative content of a certain impurity at a certain sampling time, it is obtained by calculating the sum of the instantaneous contents of all nodes in the edges of that impurity in the metabolic network at that sampling time.
[0049] S25, calculate the content of each impurity at each sampling time based on the first and second contents of each impurity at each sampling time, and construct multiple impurity information at each sampling time from the type and content of each impurity at each sampling time.
[0050] Specifically, when calculating the content of each impurity at each sampling time based on the first and second contents of each impurity at each sampling time, it is achieved through weighted calculation, where the first and second contents of each impurity at each sampling time correspond to the first weight coefficient and the second weight coefficient (preferred values are 0.8 and 0.2, respectively).
[0051] The embodiments of the present invention do not simply take the first content of each impurity obtained by liquid phase detection signal as the impurity content, but combine the first content and the second content calculated by constructing a metabolic network to obtain the content of each impurity at each sampling time, thereby improving the accuracy of impurity content calculation and providing accurate basic data for subsequent adjustment of catalytic control parameter vector.
[0052] S3 determines the catalytic stage at the current sampling time based on various impurity information at each sampling time.
[0053] Specifically, the catalytic phase in this embodiment of the invention includes the primary metabolic phase, the metabolic transition phase, and the stimulatory metabolic phase.
[0054] In one specific embodiment, S3 includes: S31, calculate the proportion of each impurity at each sampling time based on the multiple impurity information at each sampling time.
[0055] Specifically, the percentage of each impurity refers to the proportion of each impurity among all impurities at the same sampling time. When calculating the percentage of a certain impurity at a certain sampling time, firstly, the total content of all impurities at that sampling time is obtained by summing the contents of each impurity in the impurity information at that sampling time. Then, the percentage of that impurity at that sampling time is obtained by calculating the ratio of the content of that impurity to the total content.
[0056] S32, construct a proportion time series based on the proportion of each impurity at each sampling time.
[0057] Specifically, assuming the proportion time series consists of the proportions of the two impurities at three sampling times, the proportion time series can be expressed as: [proportion of impurity 1 at sampling time 1, proportion of impurity 2 at sampling time 1, proportion of impurity 1 at sampling time 2, proportion of impurity 2 at sampling time 2, proportion of impurity 1 at sampling time 3, proportion of impurity 2 at sampling time 3].
[0058] S33 determines multiple turning points based on the proportion time series, and determines the catalytic stage at the current sampling time based on the multiple turning points.
[0059] In one specific embodiment, S33 includes: S331 maps the percentage time series to a two-dimensional phase plane, where each element in the percentage time series is mapped to a point in the two-dimensional phase plane.
[0060] Specifically, in the two-dimensional phase plane, two dimensions are set as proportion and sampling time. Based on the proportion of each impurity in the proportion time series at each sampling time, multiple two-dimensional points for each impurity are generated. The dimensions of each two-dimensional point include sampling time and proportion. The multiple two-dimensional points for each impurity are then mapped onto the two-dimensional phase plane.
[0061] S332 connects points representing the same impurity at adjacent sampling times on a two-dimensional phase plane to form multiple continuous trajectory curves.
[0062] Specifically, each continuous trajectory curve corresponds to a type of impurity.
[0063] S333, for each point on each continuous trajectory curve, calculate its first-order difference and second-order difference, and calculate the curvature of each point on each continuous trajectory curve based on the first-order difference and second-order difference of each point.
[0064] Specifically, in this embodiment of the invention, the first derivative represents the first-order difference, and the second derivative represents the second-order difference. When calculating the first-order difference of a point on a continuous trajectory curve, the first derivative of the continuous trajectory curve at that point is calculated to obtain the first-order difference of that point. Similarly, when calculating the second-order difference of that point, the second derivative of the continuous trajectory curve at that point is calculated to obtain the second-order difference of that point. Based on the first-order difference of that point... and second-order difference The formula for calculating the curvature k at a point on the continuous trajectory curve is as follows: .
[0065] S334: Based on the curvature of each point on each continuous trajectory curve, obtain the points corresponding to the local curvature maxima on multiple continuous trajectory curves as multiple inflection points, and filter multiple valid inflection points from multiple inflection points according to the curvature threshold.
[0066] Specifically, when obtaining the point corresponding to a local curvature maximum on a continuous trajectory curve, the continuous trajectory curve is divided into multiple segmented curves according to a preset time interval. The point corresponding to the maximum curvature in each segmented curve is taken as the inflection point of that segmented curve. Among all the inflection points of all continuous trajectory curves, multiple inflection points with curvature greater than a curvature threshold are taken as multiple valid inflection points.
[0067] S335 divides multiple effective inflection points into multiple catalytic stages according to their curvature and determines the catalytic stage at the current sampling moment.
[0068] Specifically, if a preset number of valid inflection points exist at the same sampling time, then that sampling time is determined to be a catalytic stage point. Based on the above steps, multiple catalytic stage points are obtained. According to the preset stage start curvature and preset stage end curvature corresponding to the three catalytic stages of primary metabolism, metabolic inflection, and stimulation metabolism, the multiple catalytic stage points are divided into catalytic nodes (catalytic stage start and catalytic stage end; generally, the catalytic stage at the current sampling time does not have a catalytic stage end) corresponding to the three catalytic stages of primary metabolism, metabolic inflection, and stimulation metabolism. The stage at which the catalytic stage start closest to the current sampling time is located is taken as the catalytic stage at the current sampling time.
[0069] The embodiments of the present invention construct a continuous trajectory curve through a two-dimensional phase plane. Compared with the one-dimensional impurity ratio, the two-dimensional (ratio and time) continuous trajectory curve can not only describe the change of each impurity over time (i.e., the sampling time), but also accurately identify the turning point of each impurity, that is, accurately identify the catalytic stage at the current sampling time.
[0070] S4. Based on the current catalytic stage and the various raw parameter data at each sampling time, select multiple key parameters from all raw parameters and obtain the multiple key parameter data for each sampling time for each key parameter.
[0071] In one specific embodiment, S4 includes: S41, construct the main reaction pathway at the current sampling time based on the catalytic stage at the current sampling time, wherein the main reaction pathway includes multiple directed edges and their flux weights.
[0072] Specifically, in this embodiment of the invention, the main reaction pathway refers to the core reaction network with the highest metabolic flux at the current sampling moment during the catalytic stage of penicillin V potassium enzyme catalysis.
[0073] In one specific embodiment, S41 includes: S411, construct a set of metabolites based on the impurity types in the various impurity information at the current sampling time, and construct a set of reaction nodes based on the catalytic stage at the current sampling time.
[0074] Specifically, the reaction node set is a collection of multiple reaction nodes. In this embodiment of the invention, a reaction node represents a node consisting of a catalytic stage, a main reaction type, and a substrate. Each reaction node is pre-set. For example, two reaction nodes may be in the form of: [stimulation metabolism phase, hydrolysis reaction node, substrate is potassium penicillin V], [primary metabolism phase, oxidation reaction node, substrate is methyl phenoxyacetate]. The metabolite set is a collection of various impurity types at the current sampling time, with one impurity type corresponding to one metabolic node.
[0075] S412 connects each reaction node in the set of reaction nodes to each metabolic node in the set of metabolites, forming each directed edge.
[0076] Specifically, in this embodiment of the invention, each directed edge locks the reaction source of each metabolic node, which facilitates subsequent quantification of the importance ratio of each impurity in the catalytic reaction at the current sampling time (reflected in the flux weight of each directed edge). A directed edge can be represented as: [stimulation metabolic period, hydrolysis reaction node, substrate is penicillin V potassium] → [impurity type 1 at the current sampling time].
[0077] S413, calculate the flux weight of each directed edge based on the impurity content in the various impurity information at the current sampling time and the catalytic stage at the current sampling time.
[0078] Specifically, the weight of the corresponding catalytic stage is obtained based on the catalytic stage at the current sampling time. When calculating the flux weight of a directed edge, first calculate the flux factor e of that directed edge, using the formula: Where z represents the proportion of impurities of the impurity type in the directed edge at the current sampling time. This represents the weight coefficient corresponding to the main reaction type of the reaction node in the directed edge. Then, the flux factors of all directed edges are normalized to obtain the flux weight of each directed edge.
[0079] S414 is the main reaction path at the current sampling time, which is composed of each directed edge and its flux weight.
[0080] S42, select multiple main edges from all directed edges based on the flux weight of each directed edge and the preset activity threshold.
[0081] Specifically, multiple directed edges that are greater than or equal to a preset activity threshold are designated as multiple main edges.
[0082] S43, for each original parameter, calculate the mutual information between each original parameter and each main edge based on the original parameter data at each sampling time, and select multiple original parameters whose mutual information is greater than a preset mutual information threshold as multiple key parameters.
[0083] Specifically, when calculating the mutual information between a given original parameter X and a given main edge Y based on the original parameter data at each sampling time, the formula is as follows: Where x represents all the original parameter data of the original parameter X from the catalytic start point to the current sampling time, and y represents the flux weight of the main edge Y. In this embodiment of the invention, the original parameter X represents the product of the average of the standard parameter data x corresponding to all original parameter data from the catalytic start point to the current sampling time and the flux weight. This represents the ratio of the total amount of all original parameter data corresponding to the original parameter X from the catalytic start point to the current sampling time to the total amount of data for multiple original parameter data.
[0084] S44, selects each key parameter from the multiple raw parameter data at each sampling time.
[0085] Specifically, first, select multiple key parameters from all the original parameters, and then obtain multiple key parameter data for each key parameter at each sampling time from the multiple original parameter data.
[0086] S5, based on the multiple key parameter data at each sampling time and the catalytic stage at the current sampling time, adjust the catalytic control parameter vector for the next sampling time until the penicillin V potassium enzyme catalysis process ends.
[0087] Specifically, the catalytic control parameters in this embodiment of the invention include: pH value, temperature, nitrogen supplementation rate, and stirring rate.
[0088] In a specific embodiment, when S5 adjusts the catalytic control parameter vector for the next sampling time based on multiple key parameter data at each sampling time and the catalytic stage at the current sampling time, it includes: S51, standardize the multiple key parameter data of each sampling moment to obtain multiple standard key parameter data of each sampling moment, and obtain the stage starting sampling moment of the catalytic stage at the current sampling moment.
[0089] Specifically, the standardization method is the same as the principle of S241, and will not be repeated here. When obtaining the stage start sampling time of the catalytic stage at the current sampling time, according to the preset stage start curvature and preset stage end curvature corresponding to the three catalytic stages of primary metabolism, metabolic transition, and stimulation metabolism, the sampling time corresponding to the preset stage start curvature of the catalytic stage at the current sampling time is obtained, and is used as the stage start sampling time of the catalytic stage at the current sampling time.
[0090] S52, construct a state transition function based on the preset response coefficient matrix corresponding to the catalytic stage at the current sampling time and various standard key parameter data for each sampling time from the stage starting sampling time to the current sampling time.
[0091] Specifically, in constructing the state transition function, this embodiment of the invention uses Kalman filtering. The state transition function is obtained by iterating through multiple standard key parameter data at each sampling time using Kalman filtering. The state transition function can be expressed as: Where J represents the preset response coefficient matrix corresponding to the catalytic stage at the current sampling time. Each catalytic stage corresponds to a preset response coefficient matrix, and the preset response coefficient matrix for each catalytic stage is pre-trained. Let represent the standard key parameter data vector at the i-th sampling time from the starting sampling time of the stage to the current sampling time (the standard key parameter data vector at the i-th sampling time is obtained by arranging the various standard key parameter data at the i-th sampling time in order), and u represent the pre-set system error vector. This represents a data vector of various standard key parameters from the starting sampling time of the stage to the (i+1)th sampling time of the current sampling time.
[0092] S53, based on the various standard key parameter data and state transition function at the current sampling time, determine the predicted key parameter data for the next sampling time.
[0093] Specifically, the standard key parameter data of the current sampling time is generated into a standard key parameter data vector of the current sampling time from multiple standard key parameter data, and then input into the state transition function to obtain the predicted key parameter data of the next sampling time.
[0094] S54 calculates the deviation vector between the standard key parameter data and the predicted key parameter data at the current sampling time.
[0095] Specifically, the deviation vector is obtained by calculating the vector difference between the vector composed of multiple standard key parameter data at the current sampling time and the vector composed of predicted key parameter data.
[0096] S55 determines the catalytic control parameter adjustment vector for the next sampling time based on the deviation vector and the pre-trained PID model.
[0097] Specifically, the deviation vector is input into the pre-trained PID model as an error term, and the pre-trained PID model outputs the catalytic control parameter adjustment vector for the next sampling time from the current sampling time.
[0098] Specifically, when the catalytic control parameter adjustment vector for the next sampling time is output through the pre-trained PID model, the calculation method is as follows: Catalytic control parameter adjustment vector for the next sampling time = pre-trained proportional coefficient × deviation vector + pre-trained integral coefficient × pre-trained error integral + pre-trained differential coefficient × pre-trained error rate of change over time.
[0099] S56. The catalytic control parameter adjustment vector at the next sampling time is superimposed on the catalytic control parameter vector at the current sampling time to obtain the catalytic control parameter vector at the next sampling time.
[0100] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.
Claims
1. A method for synthesizing penicillin V potassium via enzyme catalysis, characterized in that, include: S1, acquire multiple raw parameter data and liquid phase detection signals at each sampling moment from the catalytic start point to the current sampling moment during the penicillin V potassium enzyme catalysis process; S2, based on multiple raw parameter data and liquid phase detection signals at each sampling time, identifies multiple impurity information at each sampling time during the penicillin V potassium enzyme catalysis process; S3, determine the catalytic stage at the current sampling time based on various impurity information at each sampling time; S4. Based on the catalytic stage at the current sampling time and the various raw parameter data at each sampling time, select various key parameters from all raw parameters and obtain the various key parameter data for each sampling time for each key parameter. S5, based on the multiple key parameter data at each sampling time and the catalytic stage at the current sampling time, adjust the catalytic control parameter vector for the next sampling time until the penicillin V potassium enzyme catalysis process ends.
2. The method for synthesizing penicillin V potassium by enzyme catalysis according to claim 1, characterized in that, S2 includes: S21, convert the liquid phase detection signal at each sampling time into a chromatogram at each sampling time, and obtain multiple impurity chromatographic peaks in the chromatogram at each sampling time; S22, compare multiple impurity chromatographic peaks in the chromatogram at each sampling time with a pre-set impurity database to obtain the impurity corresponding to each impurity chromatographic peak at each sampling time; S23, calculate the peak area of the impurity chromatographic peak corresponding to each impurity at each sampling time, and calculate the first content of each impurity at each sampling time based on the peak area of the impurity chromatographic peak corresponding to each impurity at each sampling time; S24, determine the types of multiple impurities at each sampling time, and construct the metabolic network for each sampling time based on the types of multiple impurities at each sampling time and the multiple original parameter data at each sampling time, and calculate the second content of each impurity at each sampling time based on the metabolic network at each sampling time; S25, calculate the content of each impurity at each sampling time based on the first and second contents of each impurity at each sampling time, and construct multiple impurity information at each sampling time from the type and content of each impurity at each sampling time.
3. The method for synthesizing penicillin V potassium by enzyme catalysis according to claim 2, characterized in that, S24 includes: S241, standardize the multiple raw parameter data at each sampling time to obtain multiple standard parameter data at each sampling time; S242, calculates the oxygen uptake rate, carbon dioxide release rate, respiratory quotient and dynamic derivative at each sampling time based on multiple standard parameter data at each sampling time, and determines the main cause of impurity generation at each sampling time based on the oxygen uptake rate, carbon dioxide release rate, respiratory quotient and dynamic derivative at each sampling time. S243, determine the types of multiple impurities at each sampling time based on the main cause of impurity generation at each sampling time, and generate the instantaneous content of each impurity at each sampling time based on the types of multiple impurities at each sampling time and the multiple standard parameter data at each sampling time; S244, construct the metabolic network for each sampling time based on the instantaneous content of each impurity at each sampling time; S245, calculate the cumulative content of each impurity at each sampling time based on the metabolic network at each sampling time, and use the cumulative content of each impurity as the second content of each impurity at each sampling time.
4. The method for synthesizing penicillin V potassium by enzyme catalysis according to claim 3, characterized in that, The main causes of impurity generation include impurities generated by metabolic byproducts, impurities generated by oxidation, or impurities generated under normal conditions. S242 includes: S2421, calculates the oxygen uptake rate at each sampling time based on the intake flow rate data, catalytic liquid volume data, intake oxygen concentration data and exhaust oxygen concentration data from multiple standard parameter data at each sampling time. S2422, calculates the carbon dioxide release rate at each sampling time based on the intake flow rate data, catalyst volume data, intake carbon dioxide concentration data and exhaust carbon dioxide concentration data from multiple standard parameter data at each sampling time. S2423, calculate the ratio of carbon dioxide release rate to oxygen uptake rate at each sampling time to obtain the respiratory quotient at each sampling time; S2424, calculate the derivative of the respiratory quotient with respect to time at each sampling moment to obtain the dynamic derivative at each sampling moment; S2425, when the oxygen uptake rate at any sampling time is greater than a preset oxygen uptake rate threshold, the carbon dioxide release rate remains unchanged compared to the previous sampling time, the respiratory quotient decreases compared to the previous sampling time, and the dynamic derivative is negative, the main cause of impurity generation at any sampling time is determined to be impurities generated by oxidation; when the oxygen uptake rate remains unchanged compared to the previous sampling time, the carbon dioxide release rate increases compared to the previous sampling time, the respiratory quotient increases compared to the previous sampling time, and the dynamic derivative is positive, the main cause of impurity generation at any sampling time is determined to be impurities generated by metabolic byproducts; otherwise, the impurity generation at any sampling time is determined to be a normal state.
5. The method for synthesizing penicillin V potassium by enzyme catalysis according to claim 1, characterized in that, S3 includes: S31, calculate the proportion of each impurity at each sampling time based on the multiple impurity information at each sampling time; S32, construct a proportion time series based on the proportion of each impurity at each sampling time; S33 determines multiple turning points based on the proportion time series, and determines the catalytic stage at the current sampling time based on the multiple turning points.
6. The method for synthesizing penicillin V potassium by enzyme catalysis according to claim 5, characterized in that, S33 includes: S331, map the percentage time series to a two-dimensional phase plane, where each element in the percentage time series is mapped to a point in the two-dimensional phase plane; S332 connects points representing the same impurity at adjacent sampling times on a two-dimensional phase plane to form multiple continuous trajectory curves; S333, for each point on each continuous trajectory curve, calculate its first-order difference and second-order difference, and calculate the curvature of each point on each continuous trajectory curve based on the first-order difference and second-order difference of each point. S334, based on the curvature of each point on each continuous trajectory curve, obtain the points corresponding to the local curvature maxima on multiple continuous trajectory curves as multiple inflection points, and filter multiple valid inflection points from multiple inflection points according to the curvature threshold; S335 divides multiple effective inflection points into multiple catalytic stages according to their curvature and determines the catalytic stage at the current sampling moment.
7. The method for synthesizing penicillin V potassium by enzyme catalysis according to claim 1, characterized in that, S4 includes: S41, construct the main reaction pathway at the current sampling time based on the catalytic stage at the current sampling time, wherein the main reaction pathway includes multiple directed edges and their flux weights; S42, select multiple main edges from all directed edges based on the flux weight of each directed edge and the preset activity threshold; S43, For each original parameter, calculate the mutual information between each original parameter and each main edge based on the original parameter data of each original parameter at each sampling time, and select multiple original parameters whose mutual information is greater than the preset mutual information threshold as multiple key parameters; S44, selects each key parameter from the multiple raw parameter data at each sampling time.
8. The method for synthesizing penicillin V potassium by enzyme catalysis according to claim 7, characterized in that, S41 includes: S411, construct a set of metabolites based on the impurity types in the various impurity information at the current sampling time, and construct a set of reaction nodes based on the catalytic stage at the current sampling time; S412 connects each reaction node in the set of reaction nodes to each metabolic node in the set of metabolites, forming each directed edge; S413, calculate the flux weight of each directed edge based on the impurity content in the various impurity information at the current sampling time and the catalytic stage at the current sampling time; S414 is the main reaction path at the current sampling time, which is composed of each directed edge and its flux weight.
9. The method for synthesizing penicillin V potassium by enzyme catalysis according to claim 1, characterized in that, When S5 adjusts the catalytic control parameter vector for the next sampling time based on multiple key parameter data at each sampling time and the catalytic stage at the current sampling time, it includes: S51, standardize the multiple key parameter data of each sampling moment to obtain multiple standard key parameter data of each sampling moment, and obtain the starting sampling moment of the catalytic stage at the current sampling moment. S52, construct a state transition function based on the preset response coefficient matrix corresponding to the catalytic stage at the current sampling time and the various standard key parameter data of each sampling time from the stage start sampling time to the current sampling time; S53, Based on the various standard key parameter data and state transition function at the current sampling time, determine the predicted key parameter data for the next sampling time; S54, calculate the deviation vector between the standard key parameter data and the predicted key parameter data at the current sampling time; S55, determine the catalytic control parameter adjustment vector for the next sampling time based on the deviation vector and the pre-trained PID model; S56. The catalytic control parameter adjustment vector at the next sampling time is superimposed on the catalytic control parameter vector at the current sampling time to obtain the catalytic control parameter vector at the next sampling time.
