Chapter 6: Gelling properties of different soybean varieties: effects of protein subunit
6.3.5 The yield of soymilk and the properties of tofu
As shown in Table 6.6, the yield of soymilk varied from 242 to 255 g using a constant water to protein ratio (18:1) and did not show any significant difference among
the varieties. The hardness seems to be the most important textural property of tofu determining the consumer acceptability of the product and we observed great differences in hardness of tofu made from different soybean varieties (Table 6.6). The yield of GDL-tofu was not monitored in this study; however, the study by Min et al. (2005) showed that tofu moisture content significantly affected the hardness and yield of tofu. A strong positive correlation between tofu hardness and its dry matter (r = 0.859, p < 0.001) was observed which is in agreement with other studies focusing on GDL-tofu (Cheng et al., 2005; Shen et al., 1991).
Table 6.6: Soymilk yield and textural properties of tofu
Variable Minimum Maximum Mean
Soymilk yield [g] 242 255 248 Tofu DM [g/g] 0.157 0.284 0.207 Hardness 275 1392 541 Cohesiveness 0.293 0.671 0.522 Springiness 0.710 1.043 0.819 Chewiness 81 622 231 Gumminess 102 778 287 Resilience 0.269 0.871 0.423
Multivariate linear regression analysis was performed to determine the extent to which tofu hardness could be predicted from other properties of soybeans. In order to measure the contribution of individual variables without the regression coefficients being dependent on the underlying scale of measurements, standardized regression coefficients were calculated. All variables were standardized by subtracting the respective mean and dividing by its standard deviation. The standardized coefficients, then, represent the change in response to a change of one standard deviation in a predictor. The larger the standardized coefficient, the greater is the influence of that parameter in the predicted model. The table of the standardized coefficients (also named beta coefficients) allows a
comparison of the relative weight of the variables in the model. The greater the absolute value of a coefficient, the greater the weight of the variable in the model. When the confidence interval around the standardized coefficients includes 0, which can easily be observed on the chart, the weight of the variable in the model is not significant.
The chart of standardized coefficients (Figure 6.4) may be used to provide an indication as to which properties are important in relation to the observed hardness of tofu. From the data shown here it could be concluded that soybean bulk density, seed oil content or water uptake factor have null predictive capability of tofu hardness. On the other hand, standardized coefficients for individual explanatory characteristics showed that RVA slope start and WAC had the highest predictive capability in the presence of the other variables although they did not reach the statistically significant level of P < 0.05.
Figure 6.4: The chart of standardized coefficients indicating the extent of predictive ability of selected properties of soybeans in relation to hardness of tofu using multivariate linear regression analysis.
To demonstrate the usefulness of the observations, models were created to predict the hardness of tofu and the dry matter of tofu. The models were developed using the multiple linear regression analysis to assess the predictive ability of selected soybean properties as they are related to the hardness and dry matter of tofu. For each model, a chart with standardized residuals, and the input data for the variable to model was displayed, followed by a histogram of the standardized residuals (graphs not shown). The histogram was useful to detect outliers or a misfit of the model. Given the normality assumptions, there shouldn't be more than 95% of the standardized residuals out of the [- 1.96; 1.96] interval. The r is interpreted as the proportion of variability of the dependent variable (tofu DM or Hardness) explained by the model. Although the correlation is significant at p < 0.05 in regards to the number of observations, the major drawback of the r is that it does not take into account the number of variables used to fit the model. Therefore, the adjusted r (that takes into account the number of variables used in the model) was calculated for each regression equation. The goodness of fit of each model was considered as satisfactory when the adjusted coefficients of correlation (radj) were
statistically significant for the given sample size and the plots of predicted and observed (experimental) values showed consistency of the empirical models, which was confirmed by the lack of any prevalent trend in the graphs of residuals.
The resulting regression equations for the tofu hardness and dry matter are:
Hardness = 562.81 * WAC - 697.68; r = 0.590 Tofu DM = 0.0582 * WAC + 0.0789; r = 0.564
Tofu DM = -0.3118 - 5.826 * 10-6 * RVA visc. -1.601 * 10-5 * RVA peak time + 1.878 * 10-3 * RVA slope start; r = 0.601
Tofu DM = 2.354 + 1.216 * 10-2 * 11S/7S + 6.868 * 10-2 * WAC; r = 0.607 Hardness = -1144 + 98.10 * 11S/7S + 647.7 * WAC; r = 0.622
Hardness = -783.5 + 0.2894 * 3-g RVA visc. + 556.5 * WAC; r = 0.593
It should be pointed out, however, that these developed models are considered as theoretical examples because these results are based on a limited data set and require further verification. Due to differences in the gelation properties of soybean storage protein fractions, many researchers have attempted to correlate these proteins with tofu quality, but results have differed greatly. Saio (1979) reported the 11S/7S ratio in soy milk strongly affected the textural properties of tofu. He found 11S protein tofu was harder than 7S tofu because the free sulfhydryl group in 11S tofu was higher than that in 7S tofu. Taira (1990) reported the protein to lipid ratio affected the hardness of tofu, but the 11S/7S ratio did not correlate with the yield and hardness of tofu. Results from Mujoo et al. (2003) showed that the 11S fraction proteins appeared to affect tofu yield; however, no relationship between storage protein fractions and tofu firmness was observed.
Cai and Chang (1999) showed that the contribution of soybean storage proteins (glycinin and β-conglycinin) to tofu yield, hardness, and sensory quality depended on the processing method used. Different processing procedures (including making different types of tofu) may account for the discrepancies in reporting relationships between soy protein content and tofu yield/texture among the studies of Shen et al. (1991), Schaefer and Love (1992), Wang et al. (1983), and Murphy et al. (1997). The study also indicated that the controversy on the relationship of 11S and 7S proteins and their ratio to tofu texture (hardness) could be due to the different processing methods used among researchers (Saio et al., 1969; Saio, 1979; Skurray et al., 1980; Taira, 1990; Murphy et al., 1997).
6.4 Conclusions
The RVA was found useful for indicating the soybean processability for tofu. While the RVA itself is unable to provide a definite answer about the quality of soybean protein, we suggest that the RVA curve is able to reveal supplementary information in addition to a simple protein content which is currently used by many soy food processors. The traditional approach of choosing the soybean lots based on the seed size or water uptake factor does not seem to be based on scientific evidence. Nonetheless, our results suggest that rather than predicting the quality of tofu from a single characteristic of soybeans (such as total soybean protein content or 11S/7S ratio) it may be useful to combine several characteristics and the most useful appeared to be the combination of protein content, RVA parameters, 11S/7S ratio and WAC.
From the results presented in this work it can be concluded that large differences exist in soybean seed characteristics and their contributions towards the properties of the final product. The aim of this work was to investigate the relative importance of individual soybean seed traits to the functional and textural properties of soy products and the interrelationship among them. Although soy protein gels have been well investigated and it was shown that the functional properties of the processed soy proteins depend on the molecular structure of the protein, there are still many aspects that remain unclear. A bench scale tofu production test still seems to be the most appropriate for evaluating soybean varieties for suitability for tofu making. It would be useful to analyze a large set of food- grade soybean seed samples from various sources for the characteristics assessed in this study in order to confirm our results, as valuable information to food technologists.