CHEN Bing-yu, LI Qi-zhai, HU Hui, MENG Shi, Faisal SHAH, WANG Qiang, LIU Hong-zhi

1 Institute of Food Science and Technology, Chinese Academy of Agricultural Sciences, Beijing 100193, P.R.China

2 Academy of Mathematics and Systems Science, Chinese Academy of Sciences, Beijing 100190, P.R.China

Abstract Peanut protein is easily digested and absorbed by the human body, and peanut tofu does not contain flatulence factors and beany flour. However, at present, there is no industrial preparation process of peanut tofu, whereas the quality of tofu prepared by different peanut varieties is quite different. This study established an industrial feasible production process of peanut tofu and optimized the key process that regulates its quality. Compared with the existing method, the production time is reduced by 53.80%, therefore the daily production output is increased by 183.33%. The chemical properties of 26 peanut varieties and the quality characteristics of tofu prepared from these 26 varieties were determined. The peanut varieties were classified based on the quality characteristics of tofu using the hierarchical cluster analysis (HCA) method,out of which 7 varieties were screened out which were suitable for preparing peanut tofu. An evaluation standard was founded based on peanut tofu qualities. Six chemical trait indexes were correlated with peanut tofu qualities (P＜0.05). A logistic regressive model was developed to predict suitable peanut varieties and this prediction model was verified. This study may help broaden the peanut protein utilization, and provide guidance for breeding experts to select certain varieties for product specific cultivation of peanut.

Keywords: peanut tofu, process, quality, peanut variety selection, evaluation standard, prediction model

1. Introduction

The global population is expected to grow to 9.7 billion by 2050 (UN 2013). Assuming diet and income inequality levels nowadays, there will be 1.6% (148.4 million) people who may face the predicament of insufficient protein intake(Medek et al. 2017). Thus, plants will play an important source to raise nutrition at a minimum cost (Joshi and Kumar 2015). Peanut is an important crop, and a widely available key source of oil and protein globally (Gong et al. 2018).China is the country with the largest peanut production and the second largest planting area in the world. Peanut has a higher nutritional value with 24-36% protein content and eight essential amino acids. Peanut's biological value is 59 and its pure digestibility can reach up to 90%which indicates that peanut protein is a valuable potential resource. Compared with soy protein, peanut protein can be easily absorbed. The content of indigestible carbohydrate in peanuts is only one-seventh of that in soybeans, so it does not cause flatulence. The white color, no flatulence factor and no beany flavor, help to overcome the defects of soybean protein quality, making peanut protein a potential protein-based substitute (Zhang et al. 2018).

Though tofu originates in East Asia and is considered as a traditional food in China, it is also becoming popular in the western countries for its nutritional value (Langthaler 2018).Since entering the western market in the 1960s, tofu became a staple food for vegetarians and an important source of protein. The saturated fat content in tofu is lower than in animal protein food, and the tofu does not contain heme and cholesterol. These advantages may help in reducing the risk of related diseases. Tofu is a better choice for all those concerned with health problems associated with excessive intake of fat, salt, and cholesterol (Craig 2009). Such researches have increased interest in the development of new foods using pulses, especially in North America (Boye et al. 2010). But the beany flavor is not widely accepted by western countries, soybean products can cause flatulence which is also a problem that plagues consumers (Zhang et al. 2010).

Our team has already established a new peanut tofu processing method by combining high-temperature pressure cooking (115°C, 0.17 MPa) and transglutaminase (TG)crosslinking (Guo et al. 2018). Guo et al. (2018) pioneered this method of peanut processing which has enriched the variety of tofu. At the same time, the whole processing takes about 3.5 h which is about 2.33 times the soybeans processing time. Also, the existing tofu production equipment cannot meet the production requirements of 115°C, 0.17 MPa. Considering the production cost and existing equipment, this method is not suitable for factory production, thus, the peanut tofu process still requires further improvement to reduce energy consumption and shorten production time. Moreover, the raw materials have a certain influence on the quality and yield of tofu, for example, the difference between soybean varieties has a significant role in determining the quality and yield of tofu,which is an important factor influencing acceptability of tofu by consumers and producers (Mujoo et al. 2003; Mao et al.2014; Meng et al. 2016). However, the effect of different peanut varieties on peanut tofu has not been reported yet.Peanut shares a wide range of cultivars throughout the globe, and different cultivars possess a varied range of compositional properties due to genetic or environmental variance. Moreover, different raw materials (cultivars)vary in their composition characteristics at the molecular level, which influences the properties and qualities of the desired products (Wang et al. 2017; Shah et al. 2019). At present, there is no evaluation method for the suitability of peanut processing tofu, and peanut protein cannot be fully utilized. There's a requirement of identifying the suitable peanut varieties for peanut tofu processing and improving the utilization of peanut protein.

The objectives of this article were: (1) to establish the feasible production process of peanut tofu at an industrial scale; (2) to screen the peanut varieties suitable for processing peanut tofu; and (3) to establish the evaluation standard of peanut tofu and a prediction model of predicting suitable peanut varieties for peanut tofu processing.

2. Materials and methods

2.1. Materials

Thirty-one peanut cultivars from China were selected according to the planting area, annual output, gelation for this study. Twenty-six peanut varieties (CN01-26) were used for the construction of the prediction model, and five varieties (V1-V5) were used to test the accuracy of the prediction model. All varieties were provided by the local academy of agricultural sciences.

TG (1 000 U g-1) was bought from Pangbo Enzyme Co.,Ltd. (Nanning, China). MgCl2(Food Grade) was bought from Angel Yeast Co., Ltd. (Yichang, China). CaSO4(Food Grade) was bought from Hubei Meiji Gypsum Products Co., Ltd., China. Glucono-δ-lactone (GDL) (Food Grade)was bought from Shanghai Xinluoluo Food Co., Ltd. China.

Two commercially available soybean tofu products(commercial tofu): CaSO4tofu (southern tofu) and MgCl2tofu (northern tofu) of Baiyu Tofu (Beijing Soybean Industry Corporation) which has the largest market share in China,were used for the texture comparison.

2.2. Stepwise method for peanut tofu preparation

Preparation of peanut cakeThe processing was based on the peanut oil process with minor adjustments. After roasting at 80°C for 20 min, peanut kernels were peeled by hot air. Pressing of peanut seeds with hydraulic pressing(28-32 MPa) for different time (3, 5, 15, 30, 60, and 180 min)was done to produce peanut cakes with different oil contents.

Preparation of raw peanut milkA total of 200 g of each kind of peanut cakes were soaked in tap water at room temperature for 40 min. The soaked water was discarded and certain amount of fresh water was added to grind the peanut cake by a grinder (Luoke Machinery Company, Beijing, China)into raw peanut milk. The raw peanut milk was then filtered automatically through a 120-mesh sieve. The ratio of peanut cake to liquid was 1:5, 1:6, 1:7, 1:8, 1:9 and 1:10, respectively.

Homogenization treatment The raw peanut milk was homogenized by an homogenizer (Donghua Machinery Company, Shanghai, China) with 30 MPa for 0, 1, 2, 3, 4,and 5 times, respectively.

TG treatmentTG was dissolved into homogenized peanut milk under a certain range of solution parameters (0, 0.10,0.15, 0.20, 0.25, and 0.30 mg TG 100 g-1peanut cake) at 40°C for a while (0, 5, 10, 15, and 20 min) with a stirring speed of 120 r min-1.

Coagulant and press After TG treatment, the peanut milk was heated at 100°C for 20 min and then cooled to 80°C.Different kinds of coagulants: CaSO4(1.5 g 100 g-1peanut cake), MgCl2(1.5 g 100 g-1peanut cake), and mix coagulant(1.125 g CaSO4+0.375 g GDL 100 g-1peanut cake) were used to produce tofu. Curdling process was carried by holding the coagulation time for 20 min to obtain the peanut curd.

The next process of pressing the tofu was modeled according to the method of Meng et al. (2016). Briefly,peanut curd was cut into dices, then the dices covered by gauze were put into a steel mold. Pressing procedure followed the method of Kao et al. (2003) with slight modifications. The pressing machine (Luoke Machinery Company, Beijing, China) was used to press the curd and the condition of three-step pressing was 0.1 MPa for 5 min,0.2 MPa for 10 min, and 0.3 MPa for 15 min. The tofu was weighted and stored at 4°C until further analysis. Tofu yield was calculated as the weight (g) of fresh tofu per 100 g peanut cake.

2.3. Determination of the texture of tofu

The texture of the tofu was determined following the procedure of Guo et al. (2018) with minor modifications,briefly, it was determined by a P36 aluminum cylindrical probe of the TA-TX2i texture analyzer (Stable Micro System Ltd., Godalming, UK). Instrument conditions: the trigger force was 10 g, pre-test speed, test speed, and post-test speed were 1.0, 1.0, and 10 mm s-1, respectively. The samples were cut from the central part of tofu. The software Texture Expert 1.22 which was inbuilt with the instrument was used to calculate hardness, springiness, and chewiness automatically. Each sample was tested at least eight times.

2.4. Water holding capacity (WHC) of tofu

The WHC of tofu was tested following the method of Guo et al. (2018). A total of 3 g (accurate to 0.001 g) of the sample was weighed (m0in a 10-mL centrifugal tube (with some degreased cotton at the bottom) and centrifuged for 10 min at 1 000 r min-1. Then it was dried at 105°C until the weight(m1) became constant. The WHC of tofu was calculated according to the formula: WHC (%)=(m0-m1)/m0×100.

2.5. Protein subunits analysis

Before the SDS-page experiment, firstly, peanut protein isolates (PPI) were extracted from peanut cakes of different varieties. The method of extraction followed He et al. (2014).SDS-page experiments were performed according to Feng et al. (2014) with minor modifications. The discontinuous system consisted of 5% (w/v) acrylamide stacking gel(pH 6.8) and 13% (w/v) acrylamide resolving gel (pH 8.8).The PPI was diluted to 2 mg mL-1with sample buffer,mixed well and then heated in boiling water for 5 min. After centrifugation at 6 000 r min-1for 10 min, 10 μL of supernatant was loaded into the gel and commercial molecular weight markers (11-180 kDa) (Solarbio, Beijing, China) were used to track the protein subunits. Electrophoresis was performed in the concentrated glue at 80 V for the first 45 min followed by 110 V until dye reached the bottom of the glass cassette.Finally, the gel was stained with 0.1% (w/v) Coomassie brilliant blue in 10% acetic acid and 45% fermentation alcohol solution for 50 min and destained in 10% (v/v) acetic acid (fermentation alcohol:acetic acid:water ratio=2:1:17)for overnight.

AlphaEase FC Gel Imaging System was used to capture protein electrophoresis spectrum images, and the Image Studio Lite Software was used to perform optical density determination on the electrophoresis images and quantitative analysis of different protein bands. The band purity calculation method: selected band light density/total optical density of all bands.

2.6. Determination of chemical traits

The protein content in peanuts was determined according to Kjeldahl method and the nitrogen to protein conversion factor was 5.47 (AOAC International 2012; Method 955.04;He et al. 2014). The fat content in peanuts was measured by Soxhlet extraction (AOAC International 2012; Method 920.85).

The contents of amino acids in peanuts were determined by an A300 Auto Amino Acid Analyzer (Membra Pure,Bodenheim, Germany) (Zheng et al. 2015).

The fatty acid compositions in peanuts were measured by gas chromatography using the fatty acid methylester method (Delmonte 2016).

2.7. Evaluation standard construction and establishment of prediction model

Evaluation standard constructionThe tofu prepared from 26 peanut varieties were clustered by HCA based on tofu hardness, springiness, WHC, and yield. According to the results from HCA, peanut tofu was divided into three clusters. And the best cluster based on superior tofu quality among three clusters was identified as the standard of good peanut tofu, which was then used for further evaluation.Principle component analysis (PCA) was used to determine the weight of each factor in the quality of peanut tofu.

The peanut tofu evaluation standard was constructed based on the tofu quality cluster result obtained by HCA and the quality weights obtained by PCA. The above analyses about HCA and PCA were performed on the SPSS 24.0 Software.

Prediction model constructionThe correlation tests were determined between tofu yield, WHC, hardness,springiness, and 39 chemical traits, and the significant test was determined to select the variables (P＜0.05). The selected variables were then extended and the logistic regression model was constructed. The logistic regression model was used to predict edible qualities of different peanut varieties based on their chemical properties.

Prediction model verificationFive different peanut varieties were selected randomly to test the accuracy of the prediction model. The peanut chemical properties were determined, and the results were introduced to the prediction model to predict the overall quality of peanut tofu. Also, 5 varieties were used to prepare the peanut tofu, and their yield, WHC, and texture were determined.

2.8. Statistical analysis

All experiments were performed more than thrice. A software SPSS 24.0 was used for data analysis and system aggregation class analysis. P-value＜0.05 was considered statistically significant.

3. Results and discussion

3.1. Production process optimization of peanut tofu

Effect of press time of peanut cake on the yield and texture of peanut tofuThe oil content in peanuts can reach more than 50%. The high oil content makes the preparation of tofu difficult and the taste greasy. This is one of the technical bottlenecks in preparing tofu products using peanuts as raw materials. The low-temperature pressing process removes part of the fat in the peanuts,and can obtain peanut cakes with lower oil contents, which overcomes the shortcoming caused by the high fat contents in the raw materials. Therefore, different pressing times were set to obtain peanut cake with different oil contents, and the residual oil contents in the peanut cake were measured:3 min (25.57%), 5 min (21.57%), 15 min (17.55%), 30 min(13.24%), 60 min (9.07%), and 180 min (4.79%). The peanut cake with different oil contents was used as raw material to prepare peanut tofu. As shown in Fig. 1-A and B, the yield of tofu increased to 180.33 g 100 g-1peanut cake first,then became stable when pressing time was 5 to 30 min and finally increased with the increase in press time. The yields were similar when press time was 5 to 30 min with no significant difference. When the press time was 5 min with 21.57% oil content, the peanut tofu had a better texture,the hardness was significantly lower than that of the MgCl2tofu, and there was no significant difference in springiness between the sample tofu and the commercial tofu. The chewiness was significantly higher than that of the CaSO4tofu, and there was no significant difference with the MgCl2tofu. Under the premise of guaranteeing the yield, based on the principle of saving production time and combining the texture of tofu, the pressing process was selected for 5 min as the best process.

Effect of peanut cake to liquid ratio on the yield and texture of peanut tofuAs shown in Fig. 1-C and D, the yield of peanut tofu increased gradually, peaking at 339.10 g 100 g-1peanut cake when the peanut cake to liquid ratio was 1:9.Furthermore, the springiness was significantly higher than both of MgCl2tofu and CaSO4tofu. When the peanut cake to liquid ratio was 1:9, peanut tofu had the highest yield as well as good texture and also saved production cost. Therefore,the ratio of 1:9 was the optimum condition.

Effect of homogenization times on the yield and texture of peanut tofuUnder low pressure homogeneous conditions, the hydrophobic groups inside the protein are exposed, and the molecular structure is gradually unfolded.The hydrophobic interaction promotes the formation of gel network structure and can improve the gel properties of the protein (Lü et al. 2019). At present in factory production,the commonly used homogenization pressure is less than 50 MPa, and the excessively high homogenization pressure does not conform to the current production situation, and 30 MPa is a commonly used homogenization pressure in industrial production.

As shown in Fig. 2-A and B, the yield of tofu grew dramatically when homogenized once, twice, and thrice.After reaching the peak at 339.10 g 100 g-1peanut cake when homogenized thrice, the yield decreased with the increase in homogenization times. The highest yield was 339.10 g 100 g-1peanut cake when it was homogenized thrice, and the hardness, chewiness and springiness of peanut tofu were between MgCl2tofu and CaSO4tofu. The yield of tofu was significantly increased around 40 g 100 g-1peanut cake after homogenizing three times. On the premise of guaranteeing the yield, homogenization process for three times was chosen as the best process. The texture of tofu will be improved in the later steps.

Fig. 1 Effect of press time and peanut cake to liquid ratio on the yield and texture of peanut tofu. Hardness of CaSO4 and MgCl2 tofu, 610.56 and 300.00 g, respectively; springiness of CaSO4 and MgCl2 tofu, 96.60 and 94.15%, respectively; chewiness of CaSO4 and MgCl2 tofu, 234.87 and 461.45 g, respectively.

Fig. 2 Effect of homogenization times and transglutaminase amount on the yield and texture of peanut tofu. Hardness of CaSO4 and MgCl2 tofu, 610.56 and 300.00 g, respectively; springiness of CaSO4 and MgCl2 tofu, 96.60 and 94.15%, respectively; chewiness of CaSO4 and MgCl2 tofu, 234.87 and 461.45 g, respectively.

Effect of TG treatment on the yield and texture of peanut tofuAs illustrated in Fig. 2-C and D, the yield of tofu raised with the increase in TG concentration, and the highest yield was 339.10 g 100 g-1peanut cake when the TG addition amount was 2.5‰. The hardness and chewiness of peanut tofu were significantly lower than that of commercial tofu.There was no significant difference in the springiness between sample tofu and commercial tofu. At the same time, when the TG addition amount was 3.0 and 2.5‰,there was no significant difference in the yield and texture of tofu. Based on the principle of saving production costs,2.5‰ TG concentration was the best process considering textural prope

r

ties and tofu yield.

In respect to TG treatment time, as shown in Fig. 3-A and B, the yield of tofu grew dramatically, and then it got stabilized with the increase in TG treatment time. When the TG treatment time was more than 5 min, the yields were higher than 300 g 100 g-1peanut cake for each, and there was no significant difference. When the TG treatment time was 10 min, the hardness and chewiness of the sample tofu had no significant difference with CaSO4tofu, and the springiness was significantly higher than that of the commercial tofu. When the TG treatment time was 10 min,the tofu had both high yield and good texture, so 10 min was selected as the optimal treatment time condition.

Effect of coagulant on the yield and texture of peanut tofuAs shown in Fig. 3-C and D, with mix coagulant(1.125% CaSO4+0.375% GDL), peanut tofu showed superior texture characteristics, its hardness and chewiness were higher than the commercial tofu and guaranteed higher yield(310.75 g 100 g-1peanut cake), therefore the mix coagulant(1.125% CaSO4+0.375% GDL) was selected as the best process condition among the studied coagulant.

Fig. 3 Effect of transglutaminase treatment time and coagulant type on the yield and texture of peanut tofu. Hardness of CaSO4 and MgCl2 tofu, 610.56 and 300.00 g, respectively; springiness of CaSO4 and MgCl2 tofu, 96.60 and 94.15%, respectively; chewiness of CaSO4 and MgCl2 tofu, 234.87 and 461.45 g, respectively.

According to the experimental results above, the optimal conditions for peanut tofu production were confirmed.Firstly, the optimal pressing time for peanut kernels was 5 min and the optimal ratio of peanut cake to liquid was 1:9(w/v). Secondly, peanut milk was homogenized 3 times and treated with 2.5‰ TG for 10 min. Then, after the boil process, 1.125% CaSO4and 0.375% GDL were added to the boiled peanut milk to coagulate. Finally, after curdling for 20 min, the peanut curd was cut into dices and pressed to obtain the peanut tofu.

3.2. Chemical properties of different peanut varieties

The basic chemical properties of 26 peanut varieties were studied and shown in Table 1. The result illustrated that the coefficient of variation (C.V.) of three indicators: crude protein,fat, 23.5-kDa protein band were less than 10%, indicating the differences of these three indicators among the varieties were slightly varied. The C.V. of the other indicators were higher which illustrated that there was a great difference that exists among chemical traits of 26 varieties.

Compared with the medians and means, except for oleic acid and linoleic acid content (because the selected peanut varieties contain high oleic acid varieties), the medians were closed to the mean which indicated that fewer outlier data existed. The above analysis illustrated that the selected peanut varieties had universality and were representative.

3.3. Quality characteristics of tofu processed from different peanut varieties

In terms of yield, the yield of tofu in different peanut varieties varied widely, from 153.95 g 100 g-1in CN24 to 310.75 g 100 g-1in CN11, with a C.V. of 16.69%. The WHC ranged from 55.98% of CN02 to 72.63% of CN22, with a C.V. of 7.54%. The hardness ranged from 270.35 to 657.32 g, the C.V. was 28.63%. The range and variation of springiness were limited, and C.V. was recorded as only 0.86%. These results indicated that the tofu quality of different peanut varieties differed, and the differences in yield and hardness were more obvious.

Cluster analysis was applied to classify tofu according to its multivariate characteristics. In this study, HCA was carried out according to the apparent quality (yield, WHC)and texture quality (hardness, springiness) of tofu. As shown in Fig. 4 and Table 2, different varieties of peanut tofu were divided into three clusters and each cluster showed different characteristic quality.

The yield of peanut tofu in the first cluster and the second cluster was not significantly different, but the hardness of the second cluster of tofu was significantly lower than that of the first cluster of peanut tofu, and the WHC and springiness of the second cluster of tofu were significantly higher than those of the first cluster of tofu (P＜0.01). Unlike soybeantofu, preparing peanut tofu with an inappropriate variety may produce tofu with a rough structure and a grainy feel which appeared as over higher hardness. The hardness,springiness, WHC of the third cluster and the second cluster had no significant difference, but the yield of the second cluster was significantly higher than that of the third cluster of tofu (P＜0.01).

Table 1 Data description of chemical properties of 26 peanut varieties

Qiaoet al.(2011) studied the quality of pressed-tofu processed by different soybean cultivars. The results showed that the overall quality of tofu was negatively correlated with hardness and adhesion. Cai and Chang(1999) studied the quality of tofu produced under different agitation conditions and suggested that tofu with high yield and smooth texture is of superior quality. Based on the above studies, the quality of tofu was analyzed and ranked.

Even though the first cluster peanut tofu had high yield,but the poor WHC and the poor texture suggested that the first cluster of peanut tofu attributed the inferior quality characteristics. And the third cluster with good texture,WHC, but the low yield of peanut tofu, can be considered as the second selection choice. The second cluster had high yield, high WHC, high springiness, and relatively suitable hardness, thus, the tofu in the second cluster was identified for its superior quality. Therefore, the seven varieties in the second cluster CN01, CN07, CN11, CN15, CN18, CN22,CN25, CN26 are suitable for peanut tofu.

Table 2 Tofu qualities of different clusters

Fig. 4 The dendrogram of hierarchical cluster analysis of tofu prepared by 26 peanut varieties.

3.4. Evaluation standard construction

By studying the internal structure of the correlation coefficient matrix between variables, PCA can synthesize multiple interrelated quantitative traits into a few principal components(Keenanet al. 2012). PCA was used to study multiple numbers of tofu quality traits for weight analysis (Table 3).

On the base of HCA, the quality traits of the second cluster of peanut tofu which was evaluated as good were identified as the first grade and got the full score. While the other as the second grade for evaluation and got the half score (Table 3). Peanut tofu with a total score of 1.00 is considered to be a good-quality peanut tofu.

3.5. Correlation analysis of chemical properties and peanut tofu qualities

Correlation analysis of 39 chemical properties and the yield,WHC, hardness, and springiness of peanut tofu were carried out. Six variables were selected based on the correlation analysis (P＜0.05): arachin/conarachin; 35.5-kDa protein subunit, glutamate, polar amino acids, oleic acid, and linoleic acid.

Correlation analysis of protein subunit composition and peanut tofu qualitiesIn the study of soybean tofu,protein content in soybean grains would affect the quality of tofu (Wang 2016), but there was no similar phenomenon in peanut tofu. This may be due to the relatively small differences between peanut varieties. Wanget al. (2013)measured the protein content of 141 peanut varieties with the C.V. of protein only 7.31%. The C.V. of protein content among different peanut varieties in this study was 7.45%. It is generally accepted that the peanut protein is divided into peanut arachin (14S), peanut conarachin (8S, 2S) (Wang 2018). SDS-PAGE separates the arachin subunits (40.5,37.5, 35.5, and 23.5 kDa subunits) and the conarachin subunits (61, 18, 17, and 15.5 kDa subunits) in the peanuts(Wanget al. 2014). Different peanut varieties have different storage protein compositions. The differences in the content of arachidene and arachidene are due to genetic and environmental differences (Dawson and McIntosh 1973).As Table 1 shows, the arachin content among the varieties ranged from 38.37 to 62.30% and those of conarachin varied from 37.90 to 61.63%, arachin/conarachin ratio ranged from 0.62 to 1.64, and 35.5-kDa protein subunit ranged from 0.00 to 16.47%. The SDS-PAGE profile analysis showed C.V.larger than 10% which suggests great differences in protein composition among different peanut cultivars.

Arachin/conarachin ratio had a positive correlation with the yield of tofu (r=0.391,P＜0.05). This finding supported the result of Du (2012) that under the best gel formation conditions, the water binding capacity of arachin was significantly better than of conarachin. Protein with higher water binding capacity showed a high yield of tofu. Similar findings have been found in soy protein, according to the reports about the correlation between soybean 11S and 7S protein subunit with the tofu yield. A study found that the 11S fraction proteins (r=0.863,P＜0.01) appeared to affect the tofu yield (Mujooet al. 2003). The 11S/7S protein ratio in seed indicated genotypic influence on soybean tofu yield and hardness (r=0.91 andr=0.99, respectively;P＜0.05)(Stanojevicet al. 2011).

Content of 35.5-kDa protein subunit and the springiness of tofu reflected a positive correlation (r=0.389,P＜0.05).This finding supported the result of Fenget al. (2014) that 35.5-kDa as one of the acidic subunits in arachin was more effectively cross-linked by TG than basic subunits.Also in the soybean 11S, there is an acidic subunit A3with MW 40.74 kDa whose MW is close to 35.5 kDa (Liuet al.2007). Menget al. (2016) found a similar phenomenon that a moderate correlation (r=0.63,P＜0.1) existed between the content of A3subunit and firmness of both pressed tofu and filled tofu. The A3subunit had an important effect on the structure of tofu, possibly due to its molecular characteristics(Menget al. 2016).

Correlation analysis of amino acid and peanut tofu qualitiesThe range of glutamate among different peanut varieties was 4.81-9.55 g 100 g-1and the C.V. was 18.00%. The content of glutamate in peanut amino acids was 19.82-37.33%. Glutamate had a positive correlation with the yield and WHC of tofu (r=0.481,r=0.474,P＜0.05).TG is an enzyme that is able to catalyze the acyl-transfer reaction and form covalent cross-links between proteins(Nonakaet al. 2006). Through cross-linking the amino acid residues of protein-bound glutamine (acyl donor) and lysine(acyl receptor), TG can form both inter- and intra-molecular isopeptide bonds in proteins (Romeih and Walker 2017).TG has been used to catalyze the cross-linking of many kinds of proteins to modify the functional properties of food proteins and has been applied in tofu processing. Guoet al.(2018) found that TG is capable of forming larger particle size protein aggregated by intermolecular cross-linking and improving the texture of tofu.

The range for lysine was a low of 0.80 g 100 g-1and a high of 1.53 g 100 g-1among different peanut cultivars.Lysine is the limiting amino acid in peanuts, the content of lysine in peanut amino acids was only 3.37-4.83%. As the acyl-acceptor, if there is no sufficient lysine, water will react with glutamate, and the γ-carboxamides of glutamine residues get deamination to form glutamate residues (Andoet al. 1989) which means the low content of lysine in peanuts may cause less effect on the quality of tofu.

The range of polar amino acid content was 13.56-20.85 g 100 g-1and the C.V. among different cultivars was 11.20%.Polar amino acid and WHC had a positive correlation(r=0.410,P＜0.05). The polar amino acids are soluble amino acids that attribute for protein solubility. The highlevel of polar amino acid residues in a protein increases the protein solubility in water (Appell et al. 2018). A study has shown that the higher the concentration of soymilk soluble protein in soybean varieties, the better the quality of processed tofu was (Xu et al. 2018). However, when the company purchases raw materials, it is troublesome to prepare soymilk and measure soluble protein. But using the near-infrared as a rapid non-destructive tool to determine the content of polar amino acids can quickly determine the required indicators. At present, a highly accurate portable near-infrared peanut rapid tester has been developed,which can quickly determine the chemical indicators such as peanut protein and amino acids (Wang 2018). With this instrument, the chemical indicators in peanuts can be determined quickly.

Table 3 Peanut tofu quality evaluation standard

Correlation analysis of fatty acid and peanut tofu qualities The relationship between fatty acid in soybean seeds and the texture of tofu has been studied. Soybean varieties with higher palmitic acid tend to produce tofu with good texture, but the linoleic acid and stearic acid are not conducive (Wang et al. 2020). Similarly, in peanut tofu,the springiness showed a negative correlation (r=-0.434,P＜0.05) with linoleic acid content, while a positive correlation(r=0.440, P＜0.05) with oleic acid content. These results indicated that higher linoleic acid in peanut seeds is not beneficial to the texture of tofu, whereas the higher oleic acid content has been proved beneficial to its texture.

3.6. Prediction model construction

The logistic regression model was used to predict tofu edible quality based on the chemical traits of different peanut varieties. The response variables and explanatory variables were arranged as follows.

Response variables (Z): based on the distance between groups Pierce coefficient, where the first cluster and the third cluster were denoted as 1, and the second cluster was denoted as 2.

Explanatory variables: six variables were selected based on the correlation analysis (P＜0.05): 35.5-kDa protein subunit (%), x1; arachin/conarachin, x2; oleic acid (g 100 g-1),x3; linoleic acid (g 100 g-1), x4; glutamate (g 100 g-1), x5; and polar amino acids (g 100 g-1), x6. Then these 6 variables were expanded up to 2-order interaction as:

For each variable in Ω, we established a logistic regression model for Z marginally, sorted their P-values in an ascending order, and selected the six features with the top minimal P-values, which were:a logistic regression model was established:

This research verified the fitting accuracy, divided the fitting probability ＞0.4 into category 2, and the rest as category 1. Five out of 26 samples were misclassified(CN08, CN09, CN15, CN18, and CN24 errors), so the internal correct classification rate was 80.77%. If the number of samples increases, the classification rate and accuracy could be improved further.

3.7. Prediction model verification

Five different peanut varieties (V1-V5) were randomly selected to test the accuracy and applicability of the prediction model. These five varieties were processed into tofu, and the quality (actual value) of these five varieties was calculated by substituting the yield, WHC, hardness,and springiness into the evaluation standard (Table 4). The predicted value of each peanut variety was obtained by determining the chemical traits data (Table 4). Among the five varieties, the prediction category of V1, V2, V4, and V5was classified as prediction category 1, and the predicted results indicated that four varieties among these five were not suitable for the preparation of peanut tofu. At the same time, the tofu quality scores of these four varieties were less than 1.00, indicating that the peanut tofu qualities of these four varieties were poor. Both performances showed a consistent result with the other. The prediction results of V3and the quality of tofu made by V3indicated that this variety can produce good quality peanut tofu. The evaluation of these five varieties was performed in the process of external verification of the model, all of which were accurate with the prepared tofu characteristics.

4. Conclusion

This study established and optimized an industrial feasible processing technology of peanut tofu. Peanut cake with a fat content of 21.57% is used as the raw material. Peanut tofu can be produced by adding 9 volumes of water for grinding,homogenizing 3 times, adding 2.5‰ of TG for 10 min, and adding a mixed coagulant (1.125% CaSO4+0.375% GDL).The peanut tofu produced by this process has a high yield.At the same time, compared with the commercially available products, the hardness and chewiness of peanut tofu made by this process were significantly higher and there was no significant difference in springiness. Through variety selection, CN01, CN07, CN11, CN15, CN18, CN22, CN25,and CN26 are the peanut varieties suitable for processing peanut tofu.

Table 4 Tofu quality score and prediction category of 5 peanut varieties

Acknowledgements

This study was supported by the earmarked fund for China Agriculture Research System (CARS-13-03A), the Corps Science and Technology Development Special Promotion Achievement Transformation Guidance Plan, Xinjiang, China(2018BC012) and the TaiShan Industrial Experts Programme,China (LJNY201711).