Team:BIT-China/Model
Overview
There is no doubt that one of the basic foundations of Tasting officers and seasoning formulation is to perform accurate, quantitative estimates and simulations. Generally, the way to do this includes accurate measurements and simulations using mathematical methods.
We refer to the related work of the 2017 BIT-China team and build on it to construct new mathematical models to quantify the protein concentration changes inside the Tasting officer, as well as to visualize and quantitatively represent the entire process of the detection officer from sensing the taste substance to presenting fluorescence. Based on the signaling model, we can demonstrate that it is theoretically feasible to conduct experiments and construct Tasting officer in this effort.
By fitting and predicting the growth of laboratory-cultured Saccharomyces cerevisiae with reference to the related work of the 2017 BIT-China team, we obtained a model for the growth of engineering bacteria during the measurement process, which will help us to predict the relationship of fluorescence intensity that may be reflected by the Tasting Officer in practical applications from a macroscopic perspective.
We derived the relationship between the degree of protein hydrolysis and the properties of hydrolytic enzymes and proteins from the mechanistic level of enzymatic reaction, and established a protein enzymatic model to predict the optimal reaction time and substrate concentration for proteins to reach a certain degree of hydrolysis, which will be useful for guiding our production of our Creative Food Seasonings.
Finally, in order to formulate customized Creative Food seasonings, we needed to understand how various flavoring substances are perceived when stacked on top of each other. Therefore, we selected several flavoring substances to be used in the formulation of the seasoning and measured the Flavor intensity at different formulation ratios. Then we did the taste indicators analysis to obtain the relationship between the flavor of these substances and the formulation ratio.
We developed the model with two main aims in mind:
- Assist the wet lab team in verifying the feasibility and correctness of the experiments through mechanism-based modeling.
- Obtain patterns based on the experimental data obtained in the wet lab and to improve the model we developed
Signaling Pathway Model
The model was built based on the signaling pathway in Saccharomyces cerevisiae cells, and the mechanism was modified according to the modification of Saccharomyces cerevisiae in the wet lab. In this model we mainly focus on the concentration changes of proteins in the signaling pathway, construct a set of ODE equation models, and use MATLAB computing platform to solve them. The model can visually reflect the change pattern of fluorescence intensity produced by an engineering bacterium after sensing the taste-presenting substances.
Purpose
In our tasting officer project, we constructed a modified Saccharomyces cerevisiae as an engineered bacterium and reflected the intensity of taste by fluorescence intensity. Therefore, we hope to construct mathematical models to quantify the changes in the concentration of some proteins in the tasting officer, as well as to visualize and quantify the entire process of the tasting officer from sensing the taste substance to presenting fluorescence, so as to verify the feasibility of the effort to construct the tasting officer and to show the response of the tasting officer after binding the taste substance.
Hypothesis
In the signaling pathway model, we focused on the signaling pathway in Saccharomyces cerevisiae cells, and since there are many factors that affect the signal output, we made some assumptions about the model:
- We assume that there is no synergistic effect when the taste receptor binds to the ligand.
- We assume that the number of bound ligands is the same when binding to either the taste receptor or the pheromone receptor.
- We assume that the binding rate and initial binding concentration of the ligand is the same as that of the pheromone receptor.
- We do not consider the effect of cell growth on signaling in individual cells in the model.
- We consider the degradation of expression products during the expression of fluorescent proteins, but do not consider the leakage of promoter expression.
Model Construction
In the signaling pathway model, we mainly focus on the protein concentration changes in the pathway during the signaling process in the modified yeast cells, so we decided to use the ODE model to describe the signaling process. To show the whole process more clearly, we describe the pathway into five modules below.
Fig 1. Signal transduction pathway diagram
Activation of the Receptor
The ligand, which is the flavor substance, first binds to the receptor that senses the taste, and in the process, the conformation of the receptor is changed. The receptor that binds the ligand and is activated will activate the next module of the pathway. During the activation of the taste receptors, the taste receptors have four different states, which are shown in the figure below, along with their inter-transfer relationships.
Fig 2. Mechanism of receptor activation
The equation for this process is shown below:
T1R:T1R1/T1R3/T1R2 series receptor
The parameters of the process are listed in the following table.
| Parameter | Meaning | Value | Source |
|---|---|---|---|
| k1 | Rate constant of ligand-receptor binding | 0.0012 | Reference[1] |
| k2 | Inverse rate constant of binding ligands | 0.6 | Reference[1] |
| k3 | Rate constants that cannot be recovered after dissociation of the ligand | 0.24 | Reference[1] |
| k4 | Degradation rate constant | 0.024 | Reference[1] |
Table 1. Values of receptor activation parameters
The result of T1R receptor activation is shown in the figure below, where it can be seen that the receptor is activated in response to binding the ligand.
Fig 3. Simulation results of receptor activation
G Protein Cycle
Figure 4. G protein cycle activation process
Activated taste receptors
first turn
on the G-protein cycle, and the activated G-protein cycle produces proteins
Gβγ that will activate the next module of the response.
The
equation
of the process is shown below.
The parameters of the process are listed in the following table.
| Parameter | Meaning | Value | Source |
|---|---|---|---|
| k5 | Gβγ synthesis rate constant | 0.0036 | Reference[1] |
| k6 | Rate constant of GTP to GDP | 0.24 | Reference[1] |
| k7 | Gαβγ Synthesis rate constant | 1 | Estimated by ourselves |
| k8 | Rate constant of complex D in the next step of synthesis | 0.1 | Reference[1] |
| k9 | Dissociation rate constant of complex D | 5 | Reference[1] |
Table 2. Parameters of the G protein cycle process
The signal output curve after G protein cycle activation is shown below, and our model can convert the signal transmitted by receptor activation with basic accuracy.
Fig 5. G protein cycle activation results
MAPK Cascade Reaction
Fig 6. MAPK Cascade reaction
The MAPK cascade reaction is
activated by proteins Gβγ produced by the previous pathway and is
achieved through a series of phosphorylation processes. The final signal
output is
achieved by the activated Ste12 protein which is activated by Fus3, and the
activated Ste12 protein acts as a signal to regulate the expression of
fluorescent
proteins.
The equation of the process is shown below.
The parameters involved in the process are listed in the following table:
| Parameter | Meaning | Value | Source |
|---|---|---|---|
| k10 | Rate constants for the binding of Ste20 to Ste5 | 5 | Reference[1] |
| k11 | Rate constants for the unbinding of Ste20 and Ste5 | 1 | Reference[1] |
| k12 | The phosphorylation rate constant of Ste11 | 10 | Reference[1] |
| k13 | Ste7 bisphosphorylation rate constant | 47 | Reference[1] |
| k14 | Fus3 bisphosphorylation rate constant | 345 | Reference[1] |
| k15 | Ste5 phosphorylation rate constant | 50 | Reference[1] |
| k16 | Fus3-pp dissociation rate constant | 140 | Reference[1] |
| k17 | Fus3-pp binding rate constant | 260 | Reference[1] |
| k18 | Rate constants for the binding of Fus3-pp to Ste12 | 18 | Reference[1] |
| k19 | Fus3-pp dephosphorylation rate constant | 50 | Reference[1] |
| k20 | Rate constants for the unbinding of the double pp-Fus3 from Ste12 | 10 | Reference[1] |
| k32 | Dissociation rate constant of complex E | 5 | Reference[1] |
| k33 | Dissociation rate constant of complex F | 5 | Reference[1] |
| k34 | Dissociation rate constant of complex G | 5 | Reference[1] |
| k35 | Dissociation rate constant of complex H | 5 | Reference[1] |
| k36 | Dissociation rate constant of complex J | 5 | Reference[1] |
Table 3. Parameters of MAPK cascade reaction
The results of the MAPK pathway output signal are shown in the following figure.
Fig 7. MAPK pathway output
Expression of Fluorescent Protein
Fig 8. Fluorescent protein expression process
The MAPK pathway generates
activated
Ste12 proteins that act as regulatory factors to activate fluorescent
protein genes,
expressing fluorescent proteins that visualize the Tasting Officers binding
ligands.
Since regulation of gene expression is involved here, we decided to use the
Hill
equation to describe this process and set the relevant parameters.
The
equation
of the process is shown below.
The parameters of the process are shown below.
| Parameter | Meaning | Value | Source |
|---|---|---|---|
| k21 | mRNA expression rate | 0.382 | Team:BIT-China 2017 |
| k22 | mRNA degradation rate | 8.39 | Team:BIT-China 2017 |
| k23 | mRNA translation rate | 0.012 | Team:BIT-China 2017 |
| k24 | Mature Fluorescent protein synthesis rate | 0.0012 | Team:BIT-China 2017 |
| k25 | Mature Fluorescent protein degradation rate | 0.018 | Team:BIT-China 2017 |
| xm | Dissociation constant (concentration of ligand at which half binding is produced) | 120 | Estimated by ourselves |
| m | Hill coefficient | 1 | Estimated by ourselves |
Table 4. Values of the parameters involved in the expression of Fluorescent proteins
We can see the relationship between the concentration of fluorescent protein and time as follows.
Fig 9. Fluorescent protein concentration production versus time
Formation of the Ste5 Complex
Fig 10. The formation process of complex C
Complex C is involved in the
process
of activating the MAPK pathway, so we describe here the process of complex C
production.
The equation of the process is as follows.
The parameters of the process are as follows.
| Parameter | Meaning | Value | Source |
|---|---|---|---|
| k26 | Synthesis rate of complex A | 1 | Reference[1] |
| k27 | Degradation rate of Complex A | 3 | Reference[1] |
| k28 | Synthesis rate of complex B | 1 | Reference[1] |
| k29 | Degradation rate of complex B | 3 | Reference[1] |
| k30 | Synthesis rate of complex C | 3 | Reference[1] |
| k31 | Degradation rate of Complex C | 100 | Reference[1] |
Table 5. Parameters of complex C formation process
The concentration of complex C varies as follows.
Fig 11. Variation of the concentration of complex C
Changes in Concentration of Other Substances Involved in the Model
parameter values refer to the settings of the previous modules
We set the initial values of the parameters according to the references and the needs of the project. We got the initial values of all the parameters involved in the project listed in the following table, where the initial values of the parameters not listed are set to 0.
| Parameter | Meaning | Value | Source |
|---|---|---|---|
| sw | ligand concentration | 1000 | Reference[1] |
| T1R | Initial receptor concentration | 1666.67 | Reference[1] |
| Ste12 | Ste12 protein initial concentration | 200 | Reference[1] |
| Gαβγ | Gαβγ protein initial concentration | 1666.67 | Reference[1] |
| A | Initial concentration of complex A | 105.94 | Reference[1] |
| B | Initial concentration of complex B | 77.87 | Reference[1] |
| C | Initial concentration of complex C | 235.72 | Reference[1] |
| Ste11 | Ste11 protein initial concentration | 158.33 | Reference[1] |
| Ste5 | Ste5 protein initial concentration | 158.33 | Reference[1] |
| Ste7 | Ste7 protein initial concentration | 36.4 | Reference[1] |
| Fus3 | Fus3 protein initial concentration | 686.4 | Reference[1] |
| Ste20 | Ste20 protein initial concentration | 1000 | Reference[1] |
Table 6. Initial values of the parameters involved in the model
Conclusion
By modeling the signaling pathway, we simulated the response pathway of the Tasting Officer after binding the ligand, from which we can see that the Tasting Officer can sense the flavor substances as the ligand and respond by producing fluorescent proteins, which indicates that our efforts to build the Tasting Officer and measure the taste intensity by using the taste receptor as a sensor and coupling the signaling pathway are feasible.
Reference
[1] Kofahl B, Klipp E. Modelling the dynamics of the yeast pheromone pathway.[J]. yeast, 2004, 21(10):831.
Protein Enzymatic Model
Fig 1. Protein enzymatic process
Purpose
In our project, we use plant-derived soybean isolate protein for enzymatic digestion, and the resulting product will be the base ingredient for the configuration of the private label seasoning.
Firstly, the evaluation of the efficiency of the enzymatic hydrolysis of soybean isolate protein will be an important indicator for the production of our private ordering seasonings. Therefore, we established a relationship between the degree of protein hydrolysis and the time of Enzymatic based on the enzymatic hydrolysis reaction, and this model will establish the basis for evaluating the hydrolysis of soybean isolate proteins.
In addition, we learned that the flavor of enzymatic products is largely affected by the degree of hydrolysis of soybean protein isolate, and for specific proteases, the substrate concentration and reaction time affect the degree of hydrolysis of soybean protein isolate, which in turn affects the flavor of enzymatic products. In order to further guide the production of customized condiments and save costs, we analyze the effect of flavor on hydrolysis degree to determine the best hydrolysis degree range for flavor, and analyze the mechanism of enzymatic reaction to predict the protein hydrolysis degree under the conditions of specific enzyme, substrate concentration and reaction time, and then determine our desired enzymatic solution.
Flavor in relation to hydrolysis degree
We used different proteases and different combinations of protease ratios to hydrolyze the proteins, and studied the effect of the flavor on the proteases and the degree of hydrolysis, then analyzed the relationship between the degree of protein hydrolysis and bitterness when protein exonuclease and protein endonuclease were used for hydrolysis respectively.
Fig 2. Correlation between bitterness value and hydrolysis degree of endonuclease enzymes
For protein endonuclease, the bitterness increased gradually with the increase of hydrolysis degree above 10%, and reached the highest bitterness at about 15-18% of hydrolysis degree, and then began to decline (Figure 1); for protein exonuclease, the bitterness value increased slowly at the hydrolysis degree above 10% (Figure 2); if the concentration of the same enzyme solution was compared, the highest bitterness value was 10 after endonuclease hydrolysis for the former, and 1 after exonuclease hydrolysis for the latter.
Fig 3. Correlation between bitterness value and hydrolysis degree of the exonuclease digest
To achieve the same degree of hydrolysis, the amount of exonuclease is more than 5 times the amount of endonuclease. The relationship between bitterness value and hydrolysis degree of endonuclease and exonuclease in equal proportions is shown in Figure 3, which shows a similar trend as Figure 1, but with a lower peak and a flatter overall curve, but the amount (cost) of enzyme used is greatly reduced.
Fig 4. Equal mix of endonuclease and exonuclease
Through our experiments and analysis, three different ratios of protease enzymes were debugged and the hydrolysis interval that could achieve the best flavor was basically determined, and then three kinds of flavoring products were produced.
In order to save the production cost of flavored condiments and adjust the optimal substrate concentration and enzymatic reaction time, we took the process of trypsin digestion as an example in the next ordinary differential model, and derived the expression of protein hydrolysis degree as a function of initial protease concentration and protein concentration under the conditions of known protease related constants by analyzing the mechanism of enzymatic reaction. Using our experimental measurements of trypsin digestion, our model fits well and can therefore be used to predict the optimal reaction time and the optimal substrate concentration for the production of flavored condiments, which can then be used as a guide for subsequent production.
ODE Model
We set the parameters for the model as follows.
| Parameters | Meaning |
|---|---|
| e0 | Initial concentration of protease |
| [E] | Unbound activated protease concentration |
| s0 | Protein initial concentration |
| [S] | Protein concentration |
| [P] | Concentration of protein enzymatic digestion products |
| e | Total concentration of protease |
| h | Degree of protein enzymatic digestion |
| r | Reaction rate of hydrolysis reaction |
Table 1.Parameters of ODE model
First we observe the enzymatic reaction:
The reaction rate of the enzymatic reaction depends on the second step，the reaction rate can be described as follows.
The process of protease inactivation can be expressed as follows.
The deactivation rate:
from (1), (2) we can find
The mechanism of substrate inhibition in this process can be described as follows.
The mechanism of product inhibition in this process can be described as follows.
When the enzymatic reaction reaches equilibrium, the concentration of the enzyme-protein complexcan be set to be essentially constant according to the the Pseudo Steady State Hypothesis. So we can find
The total concentration of protease can be expressed as
Due to the Pseudo Steady State Hypothesis, the concentration of the protein [S] can be approximated as s0 and the concentration of the product [P] is approximated as p.
Combining (4)(5)(6)(7) with (8) we can derive
From (9) and (3) we can obtain that
By integrating equation (11) we can obtain that
From (2), (10) and (12), the expressions for the rate of protein enzymatic reaction under the conditions of combining substrate inhibition and product inhibition can be obtained as follows.
where the expressions for the parameters a,b are as follows.
Similarly, the forms of parameters a and b can be derived for the cases of no substrate inhibition and product inhibition, substrate inhibition only and product inhibition only. And the expressions for a and b for all cases are listed in the following table.
Table 2. Enzymatic rate expression
After considering the references and the experimental conditions, we decided to use the conditions in which only substrate inhibition is present and not product inhibition, which is more suitable for the experimental situation.
Therefore, the relationship between the hydrolysis degree and the time under the conditions of protease digestion can be expressed as follows.
In the process of enzymatic digestion with trypsin, the parameters involved are set as follows.
| Parameters | Meaning | Value | Source |
|---|---|---|---|
| km | Michaelis constant | 0.0748(g/L) | Reference [1] |
| ks | Substrate inhibition constants | 7.961(g/L) | Reference [1] |
| k2 | Rate constants of enzymatic hydrolysis reactions | 38.439/mmol | Reference [1] |
| kd(kmk3) | Enzyme inactivation constants | 9.358/mmol | Reference [1] |
| s0 | Initial concentration of protein | 2.439g/L | experimental condition |
| e0 | Initial concentration of protease | 0.099g/L | experimental condition |
Table 3. parameter values of hydrolysis degree equation
Thus we can obtain the relationship between the hydrolysis degree and time.
Result
The above equation allows us to obtain the simulation results of the degree of hydrolysis of soybean isolate by trypsin enzymatic digestion.
Fig 5. the Relationship between Protein hydrolysis degree and the time
As shown in the figure, according to the experimental record of the enzymatic digestion part, the hydrolysis degree at 8 h of the enzymatic digestion experiment is 24.85%, which is close to the predicted value of 25.162% at 8.1h (484min). It can be seen that our model fits well.
Using the above model we predicted the degree of hydrolysis of a certain concentration of protein (2.439 g/L) using different concentrations of trypsin enzymes for a fixed enzymatic digestion time (8h) as shown below.
Fig 6.Relationship between protein hydrolysis degree and enzyme concentration
By finding the relationship between protein hydrolysis and enzyme concentration, we can obtain a range of enzyme concentrations that should be added during the enzymatic digestion of soybean isolate to obtain the desired flavor while maintaining a high rate of enzymatic digestion.
Reference
[1] QI Wei, HE Zhi-min, Mechanism and Kinetic Model of Enzymatic Hydrolysis of Protein, Journal of Tianjin University, 2005.9, 9(38), 768-773.
Yeast Growth Model
Purpose
By modeling the signaling pathway, we have understood how a Tasting Officer cell reflects the flavor intensity by fluorescence after sensing the flavor substances. However, in the practical application of taste detectors, we have to consider the fluorescence effect from a macroscopic perspective.
At the same time, by the description of the signaling pathway, we can see that in order to enhance the output fluorescence intensity, we have made some modifications to the original signaling pathway of Saccharomyces cerevisiae cells by knocking out Sst2, Far1 and Ste2 genes. In the following, we call this cell the engineered yeast cell.
Therefore, we would like to build mathematical models to estimate the number of Tasting Officer cells and analyze the differences in growth between normal yeast cells and our Tasting Officer cells to determine the extent to which our modifications to the signaling pathway would have an effect on yeast cells.
Model Construction
Based on the assumption that the growth of both normal yeast cells and the Tasting Officer cells during growth is limited only by environmental space and nutritional factors, we used a Logistic growth model to describe the growth of the cells.
Refer to the 2019 BIT-China team for related ideas, the integral form of the logistic growth model can be obtained by solving for the initial value N0,initial moment t0 as
In order to determine the parameters among the model, we performed culture experiments with yeast cells and engineered yeast cells and measured the absorbance (OD600) of the yeast suspension at different moments in the culture cycle by spectrophotometer. A review of the literature showed that the values of the bacterial density of the yeast suspension y showed a linear relationship with the absorbance x in the effective interval (0.2-0.8). And that is
Therefore the indicator of absorbance reflects the concentration of the suspension at the time of measurement and thus the growth of the yeast.
Based on the information reviewed [1] we learned that the relationship between the OD of the yeast suspension and the original OD after dilution is not linear, but shows a power exponential form. Based on the references and the experimental data, we first determined the relationship between the OD600 of the original yeast solution and the OD600 value after dilution during the actual measurement at different dilution multiples as follows.
The parameters in the equation are set as follows
| Parameters | Meaning |
|---|---|
| N | OD600 values obtained from actual measurements |
| N0 | OD600 values measured for undiluted bacterial broth |
| x | Dilution times |
| k | Index Factor |
Table 1. Parameters of the relationship between the dilution times of the bacterial solution and the OD value
We measured the OD600 values of the undiluted bacterial broth and the broth diluted at the corresponding multiples, fitted them, and solved them with the software to obtain the following values of the above parameters.
| Parameters | Value | Source |
|---|---|---|
| N0(Normal yeast) | 0.114 | Experimental data |
| N0(Engineered yeast) | 0.245 | Experimental data |
| k(Normal yeast, logarithmic growth period) | 0.232 | Derivation from Experimental data |
| k(Normal yeast, stable growth period) | 0.624 | Derivation from Experimental data |
| k(Engineered yeast, logarithmic growth period) | 0.132 | Derivation from Experimental data |
| k(Engineered yeast, stable growth period) | 0.372 | Derivation from Experimental data |
Table 2. parameter values of the equation of bacterial solution dilution and OD
The growth curves of cultured yeast cells and taste detection officers were obtained by fitting the experimental data as follows.（The OD600 of the undiluted bacterial solution calculated by the above equation was used, and some of the measured OD600 values that did not fall within the 0.2-0.8 range were excluded.）
Fig1 Experimental data of yeast growth
Based on the experimental data, we fitted the parameters in the above logistic model using a nonlinear fitting method.
| Parameters | Meaning | Fitting initial value (Normal yeast) | Fitting initial value (Engineered yeast) | Value(Normal yeast) | Value(Engineered yeast) |
|---|---|---|---|---|---|
| k | Growth rate of yeast cell population | 0.20 | 0.40 | 0.2891 | 0.1204 |
| Nm | Maximum OD value | 2.00 | 3.00 | 1.7704 | 2.3544 |
Table 3.yeast growth prediction curve parameter values
We estimated the following yeast cell growth curves.
Fig 2. Predicted curve of yeast growth
It can be seen that our predicted yeast cell growth curve fits well with the actual growth curve.
Conclusion
- Engineered Saccharomyces cerevisiae has a higher growth threshold and a longer growth period.
- Normal yeast ends logarithmic growth period earlier than engineered yeast.
Reference
[1] CAO Guozhen, MIU Jianshun. Determination of Saccharomyces cerevisiae cell suspension concentration by spectrophotometry. China Brewing, 2014, 4(33), 129-133.
Taste Indicators Analysis
After the flavor substances are obtained by enzymatic digestion and the flavor intensity of the flavor substances is measured by the Tasting Officer, we have to consider the process of formulating the flavoring. Since the interaction relationship between various flavor substances is not clear, it is difficult for us to determine the ratio of flavor substances required to achieve a certain flavor. Therefore, we formulated several groups of flavorings with different ratios using several flavor substances and measured the intensity of each flavor. By analyzing the correlation between the strengths of individual flavors, we studied the pattern of flavor substance interactions to obtain flavors, which helped us to construct private custom flavorings.
Experimental Results
During the experiments we used three enzymatic products as flavoring substances, and the ratios of each group to prepare the seasoning are listed below.
| Group | a | b | c | d | e | f | g | h |
|---|---|---|---|---|---|---|---|---|
| Proportion | 1:4:1 | 1:1:4 | 4:1:1 | 3:2:1 | 2:3:1 | 1:2:3 | 2:2:2 | 0:0:6 |
Table1.Experimental ratios for each group
We measured the intensity of the various flavors presented by each group through the electronic tongue device in the following table.
| Bitterness | Sourness | Umami | Salty | Astringent | Umami aftertaste | Astringent aftertaste | Bitterness aftertaste | |
|---|---|---|---|---|---|---|---|---|
| a | 6.71 | -41.75 | 14.09 | 1.37 | 0.58 | 0.75 | 0.16 | -0.43 |
| b | 3.65 | -37.12 | 12.80 | 7.42 | 0.49 | 0.94 | 0.27 | -0.07 |
| c | 6.81 | -42.17 | 14.10 | 1.23 | 0.63 | 0.82 | 0.15 | -0.44 |
| d | 6.80 | -42.03 | 14.07 | 1.21 | 0.63 | 0.84 | 0.16 | -0.44 |
| e | 6.69 | -41.77 | 14.05 | 1.32 | 0.62 | 0.88 | 0.17 | -0.44 |
| f | 4.20 | -38.06 | 13.09 | 5.98 | 0.46 | 0.96 | 0.24 | -0.20 |
| g | 5.12 | -39.40 | 13.46 | 4.24 | 0.49 | 0.84 | 0.20 | -0.30 |
| h | 2.98 | -34.51 | 12.32 | 9.07 | 0.59 | 0.84 | 0.35 | -0.13 |
| tasteless | 0.00 | -13.00 | 0.00 | -6.00 | 0.00 | 0.00 | 0.00 | 0.00 |
| Error limit | 0.18 | 0.49 | 0.07 | 0.01 | 0.01 | 0.06 | 0.01 | 0.02 |
| Base value | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 | 0.00 |
Table2.Flavor intensity values for each group
All data are absolute output values with artificial saliva (reference solution) as the standard. The state of artificial saliva in the electronic tongue test simulates the state of the human mouth when there is only saliva.
Tasteless is the tasteless point, which is the output of the reference solution. The reference solution (reference) consists of KCL and tartaric acid as the taste value, so the tasteless point for sour taste is -13, and the tasteless point for salty taste is -6. Using this as a benchmark, when the taste value of the sample is lower than Tasteless, it means that the sample has no such taste, and vice versa, it shows the corresponding taste.
From the measurement results, it can be seen that the main taste indicators of the samples are: sour, bitter, umami and salty.
Identify Valid Taste Indicators
Fig 1. Flavor intensity values for each group
With the output of the reference solution as "0", the tasteless points of all indicators except acidity and saltiness were 0. We took the taste items greater than the tasteless points as the evaluation objects. Since the reference solution is prepared by potassium chloride and tartaric acid, the reference solution contains a small amount of acid and salt, and the tasteless points of sour and salty taste are -13 and -6 respectively.
Items below the tasteless point can be considered as tastes that are not present in this sample. As seen in the above figure, the main taste indicators of the sample are umami, bitterness and saltiness. Next, we will analyze the samples according to their main taste indicators.
Analysis of Various Taste Indicators
Bitterness
Fig 2. Bitterness intensity of different groups
The bitterness intensity of the different groups of samples can be seen in the above graph. The bitterness of the umami flavoring is different from the bitterness of the bitter samples such as coffee, which may be an expression of flavor, and the greater the bitterness the richer the taste may be.
Salty
Fig 3. Savory intensity of different groups
The salty intensity of the different groups of seasonings can be seen in the above graph. The saltiness reflects the taste presented by inorganic salts such as table salt in the samples.
Umami and Richness
Fig 4. Umami and richness of different groups
The above graph shows the intensity of umami and the richness (umami aftertaste) of the seasonings presented by different groups.
2D Bubble Diagram of the Main Taste Indicators
Fig 5 .x-axis is umami, y-axis is umami aftertaste and the bubble size indicates bitterness
Fig 6. x-axis for umami , y-axis for umami aftertaste, bubble size indicates salty taste
Figure 5 compares the umami and richness (i.e. umami aftertaste) of the samples, with the bubbles indicating the bitter ones. Figure 6 compares the umami and richness (i.e. umami aftertaste) of the samples with bubbles for the salty taste. It can be seen from the figure, the sample umami aftertaste is closer, umami from h, b, f, g, A class gradually increasing, and from the sample proportion can be seen, umami and sample 3 accounted for a positive ratio, that is, with the sample three then can be seen, a, c, d and e group is closer, and the bubble size is also closer, can be placed in a class, recorded as A class.
It can be seen from the graph, with the umami increases, bitterness increases and salty taste decreases; the four groups of data in category A are larger and have larger values than the other four groups of bubbles, and the four groups of data in category A are smaller and have smaller values than the other four groups of bubbles in terms of salty taste.
PCA Principal Component Analysis
| Taste | 1 | 2 |
|---|---|---|
| Bitterness | -0.152 | -0.065 |
| Sourness | 0.149 | 0.215 |
| Umami | -0.151 | -0.163 |
| Salty | 0.151 | 0.122 |
| Astringent | -0.1 | 0.611 |
| Umami aftertaste | 0.092 | -0.656 |
| Astringent aftertaste | 0.144 | 0.31 |
| Bitterness aftertaste | 0.15 | -0.054 |
| Eigenvalue | 6.549 | 1.006 |
| Contribution Rate | 81.862 | 12.571 |
Table3. Eigenvector
Fig 7. Scatter plot of the second principal component score against the first principal component score pair
We used factor analysis with dimensionality reduction in SPSS to perform principal component analysis for a total of eight data sets of bitterness, sourness, umami, saltiness, astringency, richness, astringent aftertaste, and bitter aftertaste for different ratios of three seasonings measured by e-tongue. We took the two principal components with the highest contribution rate, the first principal component contributed 81.86%, the second principal component contributed 12.57%, and the cumulative contribution rate of the two was 94.43%. Combined with the PCA sensor contribution table, it can be seen that the contribution to the first principal component is mainly salty, sour, astringent aftertaste and bitter aftertaste, which mainly reflect these indicators, and the larger contribution to the second principal component is astringent.
Cluster analysis was performed on the samples and the results are shown in the figure below. As can be seen from the figure, the smaller the distance between the samples, it means that the integrated taste sensation is closer, and vice versa, the samples have more differences in taste sensation. It can be seen that the four groups of samples a, e, d, and c have a closer integrated taste perception, which is the same as our conclusion for comparing umami and richness in examining bitterness and saltiness. In terms of scores, it can be seen that the first principal component and the second principal component composite scores have the same trend, i.e., samples with high first principal component composite scores also have high second principal component composite scores. For the samples we measured, the two principal component composite scores of h, b, f, g, and A category (a, e, d, and c) can be seen to have a decreasing trend by the principal component analysis method. From the first principal component representing salty, sour, astringent aftertaste and bitter aftertaste indicators, and the second principal component representing astringent indicators, it can be seen that h, b, f, g, Class A (a, e, d, c) salty, sour, astringent aftertaste and bitter aftertaste as well as astringent process have a decreasing trend, and the greater these indicators may be richer in taste for the umami-based condiments.
Comprehensive Analysis and Conclusion
From the above analysis we can conclude that:
For the above samples, the umami decreases gradually from h, b, f, g, and A category (a, e, d, and c), and the salty, sour, astringent aftertaste and bitter aftertaste as well as astringency decreases gradually. Combined with our configuration of seasoning ratios 1, 2, and 3, it can be seen that seasoning #3 has a greater effect on the overall taste, and an increase in seasoning #3 leads to an increase in umami and a decrease in taste complexity (salty, sour, astringent aftertaste and bitter aftertaste as well as astringency), while changes in the ratios of seasonings #1 and #2 have less effect on the experimental results.
Such experimental results and analysis have a guiding meaning for us to further promote our products to different groups of people, we can appropriately configure the mix ratio of seasoning 1 and 2 according to the taste preferences of the target group for seasoning 1 and 2, and rely on seasoning 3 to adjust the overall umami and complexity of the seasoning.
Flavor Card
One of the foundations of customized Creative Food seasonings is to understand what the customers want in terms of their favorite flavors. In many cases, people don't really know what they really want. Therefore, if we want to understand the needs of the customers, we need to do two things. First, we need a way to let customers know what their favorite flavors are, and then we need to get information about their favorite flavors.
Based on these two objectives, we developed a simple web application "Flavor card", which contains many common staple foods, dishes and drinks, and records the intensity of these foods corresponding to various flavors. In the process, customers can select any kind of food they like and mark the preference (integer, 1-5) for each food they choose. By analyzing the users' choices, we can get the average preference of users about each flavor, and get the user’s favorite flavor based on the average preference. Finally, after getting the results, we prepare an exclusive color for each flavor and give the corresponding analysis, enabling the user to meet with the favorite flavor by this visualization.
It must be admitted that the current "Flavor card" is still relatively primitive in terms of the variety and number of dishes and the algorithm used to analyze the user’s preferences, there are still certain shortcomings, but we hope that this program can make some contribution to understanding people's taste needs and provide a reference for future taste research. So, if you are interested, why not give it a try? Click thisand let's meet our favorite flavors together!