波长选择SCARS和偏最小二乘法建立香菇中多糖含量预测的近红外光谱模型

为实现香菇多糖含量的快速测定,利用近红外光谱漫反射技术采集了60个香菇粉末样本在12000~3800 cm-1范围内的光谱数据,利用紫外可见光谱法测定了香菇粉末样品的多糖含量。采用多种化学计量学方法,剔除掉四个异常样本后,考察了不同的光谱预处理方法以及波长选择对模型的影响,用留一交互检验法建立了偏最小二乘(PLS)模型,并用所建立的校正模型对独立预测集样本进行了预测。结果表明,当采用二阶导数及变量稳定性的竞争自适应加权抽样法(SCARS)选择的波长对光谱进行处理时,所建立的模型预测效果最佳,在隐变量数为10时,模型相关系数为0.9906,校正均方根误差(RMSEC)为0.0523 g/100 g,预测相关系数Rp=0.9781,预测均方根误差(RMSEP)=0.0577 g/100 g,该模型具有较好的预测能力,可用于香菇多糖含量的近红外光谱快速检测。     

近红外光谱技术预测猪肉的pH、嫩度和蒸煮损失

以冷却猪肉为研究对象,评价近红外光谱（NIR）技术用于肉类物理特性预测的可行性以及不同的光谱处理方法和建模方法对预测准确性的影响。试样取自排酸24h的同一批猪胴体的小里脊肉,采集4000—10000cm-1的光谱。经外部验证的偏最小二乘（PLS）模型在预测pH时表现出良好的相关性（Rc^2=0．88,Rp^2=0．80,SEC=0．08,SEP=0．084）,嫩度与蒸煮损失模型的相关性分别是Rc^2=0．50和0．57,R;=0．34和0．50。在各种光谱预处理方法中,平滑处理结合多元散射校正（MSC）或标准正态变量变换（SNV）的效果最好。     

应用近红外光谱估测小麦叶片氮含量

研究利用近红外光谱(near-infrared, NIR)和化学计量学方法估测小麦(Triticum aestivum)新鲜叶片和粉末状干叶中全氮含量的可行性, 并建立小麦叶片氮含量估测模型, 以期为小麦氮素营养的精确管理提供理论依据。以3个小麦田间试验观测资料为基础, 分别运用偏最小二乘法(partial least squares, PLS)、反向传播神经网络(back-propagation neural network, BPNN)和小波神经网络(wavelet neural network, WNN), 建立小麦叶片氮含量的鲜叶和粉末状干叶近红外光谱估测模型, 用随机选择的样品集对所建模型进行测试和检验。结果显示, 利用PLS、BPNN和WNN 3种方法构建的近红外光谱模型均能准确地估测小麦叶片氮含量, 其中基于BPNN和WNN的模型优于基于PLS的模型, 且以基于WNN的模型表现最好。对模型进行检验的结果显示, 粉末状干叶模型的预测均方根误差(RMSEP)分别为0.147、0.101和0.094, 鲜叶模型的RMSEP分别为0.216、0.175和0.169, 模型的相关系数均在0.84以上。因此, 利用近红外光谱估算小麦叶片氮素营养精确可行, 对其他作物的氮素营养估测提供了借鉴和参考。     

小波分析在人血清中葡萄糖短波近红外定量中的应用

利用小波分析对13名志愿者18个血清样品的短波近红外光谱进行去噪预处理,以血糖仪测定的血糖为参考,采用间隔偏最小二乘法（iPLS）在700nm～1060nm短波近红外波段建立血糖浓度预测模型。由相关系数（R）和预测标准差（RMSEP）对预测模型的精确度进行了评价。预测模型的相关系数为0．9654,均方根预测误差为0．2435,并和采用傅立叶变换去噪方法及iPLS建模的结果进行了比较。结果表明：小波分析预处理数据的方法能更有效地扣除噪声干扰,使模型具有更强的抗干扰能力和更高的预测精度。     

食用调和油中花生油含量的近红外光谱分析

采用偏最小二乘法(PLS)等方法建立了食用调和油中花生油含量定量分析的近红外光谱定标模型。采集食用调和油样品在4 000 cm-1~10 000 cm-1范围内的近红外漫反射光谱,光谱经一阶导数处理后,采用偏最小二乘法建立样品中花生油含量的定标模型,并用Leave-one-out内部交叉验证法对模型进行验证。模型相关系数为0.99961,校正均方根RMSEC为0.830%。比较不同光谱预处理方法对定标模型的影响,结果表明一阶导数Corr.coeff最好。采用不同的化学计量学方法建立的定标模型中以偏最小二乘回归法最理想。     

基于近红外光谱的冬小麦籽粒蛋白质含量检测

冬小麦籽粒蛋白质含量(GPC)是评价冬小麦品质的主要指标,为了研究不同建模方法对GPC检测的影响,本研究对冬小麦籽粒的近红外原始光谱进行S-G平滑、基线校正和多元散射校正等预处理,利用连续投影算法(SPA)提取冬小麦GPC的重要光谱波段,并结合偏最小二乘回归(PLSR)、主成分回归(PCR)、支持向量机(SVM)和多元线性回归(MLR)建立GPC的光谱预测模型,并综合比较模型的适用性。结果表明:经过SPA提取的特征波段为1801、1010、1109、2284、2219、2239、871、1361、1925、1849和1456 nm;模型评价方面,利用特征波段建立的SVM模型效果较好,其中校正均方根误差(RMSEC)和R2分别为0.2481和0.9760,验证均方根误差(RMSEP)和R2分别为0.3587和0.9581。研究表明,SPA+SVM预测模型在一定程度上能够实现冬小麦籽粒蛋白质的快速、无损检测。     

近红外光谱法快速测定女贞子中特女贞苷的含量北大核心CSCD

应用近红外漫反射光谱法快速测定女贞子中特女贞苷的含量。运用近红外光谱技术(NIRS)结合偏最小二乘法(PLS)建立不同产地女贞子中特女贞苷含量的定量校正模型。特女贞苷的定量校正模型内部交叉验证决定系数(R2)为0.98075,校正均方根偏差(RMSEC)为0.216,预测均方根偏差(RMSEP)为0.223,交互验证均方根偏差(RMSECV)为0.52276。该方法具有简便快速,准确无损,可用于女贞子中特女贞苷含量的快速测定。     

近红外光谱无损测定大豆种子生活力方法研究

：快速准确无损测定种子生活力是种质资源安全保存研究中的一项重要内容。采用傅立叶变换近红外漫反射光谱技术,结合偏最小二乘法,以保存不同年限的黄色大豆品种资源的种子为样品,建立其生活力的无破坏性测定数学模型,同时对不同光谱预处理方法和不同建模波段范围对模型的预测性能进行对比分析。结果表明：原始光谱在4000～10000nm全波段的模型预测精度较高。经Savitzky-Golay二介导数和标准化预处理后,生活力的PLS模型最好,校正集样品的相关系数为0.937,预测集样品的相关系数为0.902,RMSEC和RMSEP分别为2.190和2.684。可见模型预测的准确性接近常规发芽方法,能够满足种质资源快速、非破坏性活力检测的要求,为今后快速无损测定种子生活力提供了理论依据。     

近红外光谱测定小麦粉常规营养成分的模型优化

目的：应用近红外光谱（NIR）结合偏最小二乘法（PLS）建立小麦粉常规营养成分蛋白质、水分和脂肪的含量预测模型,并选择最佳模型。方法：收集117份小麦粉样品的近红外光谱,化学法测定蛋白质、水分和脂肪的含量,利用主成分分析（PCA）随机分组,81份样品用于构建模型、36份样品用作验证模型的预测能力。探讨波长范围和光谱预处理方法对所建模型预测能力的影响。结果：3个营养成分预测能力最好的模型分别是：对于蛋白质,预处理采用矢量归一化（SNV）,波长选取7 505.9~5 446.2 cm-1和4 605.4~4 242.8 cm-1,预测模型的RPD值是7.02;对于水分,无预处理,波长选择全谱12 800~3 960 cm-1,模型的RPD值是6.83;对于脂肪,无预处理,波长在9 000~4 000 cm-1,模型的RPD值是5.06。结论：近红外光谱法可以实现对小麦粉常规营养成分的快速预测,通过选择波长范围和光谱预处理方法可以显著提高模型的预测能力。     

近红外光谱法快速测定女贞子中特女贞苷的含量

应用近红外漫反射光谱法快速测定女贞子中特女贞苷的含量。运用近红外光谱技术(NIRS)结合偏最小二乘法(PLS)建立不同产地女贞子中特女贞苷含量的定量校正模型。特女贞苷的定量校正模型内部交叉验证决定系数(R2)为0.98075,校正均方根偏差(RMSEC)为0.216,预测均方根偏差(RMSEP)为0.223,交互验证均方根偏差(RMSECV)为0.52276。该方法具有简便快速,准确无损,可用于女贞子中特女贞苷含量的快速测定。     

Data fusion-based assessment of raw materials in mammalian cell culture

In mammalian cell culture producing therapeutic proteins, one of the important challenges is the use of several complex raw materials whose compositional variability is relatively high and their influences on cell culture is poorly understood. Under these circumstances, application of spectroscopic techniques combined with chemometrics can provide fast, simple, and non‐destructive ways to evaluate raw material quality, leading to more consistent cell culture performance. In this study, a comprehensive data fusion strategy of combining multiple spectroscopic techniques is investigated for the prediction of raw material quality in mammalian cell culture. To achieve this purpose, four different spectroscopic techniques of near‐infrared, Raman, 2D fluorescence, and X‐ray fluorescence spectra were employed for comprehensive characterization of soy hydrolysates which are commonly used as supplements in culture media. First, the different spectra were compared separately in terms of their prediction capability. Then, ensemble partial least squares (EPLS) was further employed by combining all of these spectral datasets in order to produce a more accurate estimation of raw material properties, and compared with other data fusion techniques. The results showed that data fusion models based on EPLS always exhibit best prediction accuracy among all the models including individual spectroscopic methods, demonstrating the synergetic effects of data fusion in characterizing the raw material quality. Biotechnol. Bioeng. 2012; 109: 2819–2828. © 2012 Wiley Periodicals, Inc.     

Diversity of Staphylococcus aureus enterotoxin types within single samples of raw milk and raw milk products

AIM: To find out if testing of up to 10 Staphylococcus aureus isolates from each sample from raw milk and raw milk products for staphylococcal enterotoxin (SE) might increase the chances of identifying potential sources of food intoxication. METHODS AND RESULTS: Altogether 386 S. aureus isolates were tested for the presence of SE by reversed passive latex agglutination (SET-RPLA), and SE genes (se) by a multiplex polymerase chain reaction (PCR). In 18 of 34 (53%) S. aureus positive samples a mixture of SE and/or se positive and negative isolates were identified. Multiplex PCR increased the number of potential SE producing strains, i.e. isolates that harboured se, with 51% among the product and 48% among the raw bovine milk isolates. Examination by pulsed-field gel electrophoresis mostly confirmed clonal similarity among isolates sharing SE/se profile, but did not further differentiate between them. CONCLUSIONS: Isolates of S. aureus collected from one sample may show great diversity in SE production and different plating media seem to suppress or favour different strains of S. aureus. SIGNIFICANCE AND IMPACT OF THE STUDY: Several isolates of S. aureus from each sample should be tested for enterotoxin production in cases with typical SE intoxication symptoms with methods that are able to reveal new SE/se.     

A potential contamination error associated with insect protein mark‐capture data

Various types of protein‐spray solutions have proven effective for externally tagging arthropods for mark‐release‐recapture and mark‐capture type dispersal research. However, there is concern that certain standardized arthropod collection methods, such as sweep netting, might lead to high incidences of protein transfer from field‐marked to unmarked arthropods during sample collection and sample handling. Native arthropods were collected in sweep nets from a field of alfalfa, Medicago sativa L. (Fabaceae). The nets also contained 10 egg white‐, 10 bovine milk‐, 10 soy milk‐, and 10 water (control)‐marked Hippodamia convergens Guérin‐Méneville (Coleoptera: Coccinellidae) that were visually distinguishable by a yellow, white, green, and blue dot, respectively. The plant debris and arthropods from each sweep net collection were then placed into either a paper or a plastic bag and frozen for storage. The contents of each sweep net sample were thawed and the color‐coded H. convergens and field‐collected arthropods were examined for the presence of each protein by an egg white (albumin), bovine milk (casein), and soy milk (soy trypsin) enzyme‐linked immunosorbent assay (ELISA). Data revealed that only 0.67, 0.81, and 0% of the field‐collected unmarked arthropods acquired an egg white, bovine milk, and soy milk mark, respectively. ELISA results also showed that all the egg white‐marked H. convergens retained their mark, but 22.1% of the bovine milk‐marked and 5.1% of the soy milk‐marked H. convergens (color‐coded beetles) lost their mark during the collection and sample handling processes.     

Determining the source of Bacillus cereus and Bacillus licheniformis isolated from raw milk, pasteurized milk and yoghurt

Aims:  Strain-specific detection of Bacillus cereus and Bacillus licheniformi s in raw and pasteurized milk, and yoghurt during processing.
Methods and Results:  Randomly selected isolates of Bacillus spp. were subjected to PCR analysis, where single primer targeting to the repetitive sequence Box elements was used to fingerprint the species. The isolates were separated into six different fingerprint patterns. The results show that isolates clustered together at about the 57% similarity level with two main groups at the 82% and 83% similarity levels, respectively. Contamination with identical strains both of B. cereus and B . licheniformis in raw and pasteurized milk was found as well as contaminated with different strains (in the case of raw milk and yoghurt/pasteurized milk and yoghurt). Several BOX types traced in processed milk samples were not discovered in the original raw milk.
Conclusions:  BOX-PCR fingerprinting is useful for characterizing Bacillus populations in a dairy environment. It can be used to confirm environmental contamination, eventually clonal transfer of Bacillus strains during the technological processing of milk.
Significance and Impact of the Study:  Despite the limited number of strains analysed, the two Bacillus species yielded adequately detectable banding profiles, permitting differentiation of bacteria at the strain level and showing their diversity throughout dairy processing.     

Enumeration of micro-organisms in processed soy products with an automated most probable number method compared with standard plate method

Aim: The automated TEMPO system (bioMerieux) is based on the most probable number (MPN) method for the enumeration of micro‐organisms in foods. In this study, we evaluated the performance of the TEMPO system as a diagnostic tool in comparison with the standard method in processed soy products. Methods and Results: A verification study was conducted using artificially contaminated soy product samples such as soy protein isolate, water‐soluble soy polysaccharides, soy milk and processed soy food. Five types of micro‐organisms were analysed using the automated MPN method (total aerobic bacteria, total coliforms, Enterobacteriaceae, yeast and mould and Staphylococcus aureus) vs the standard plate method. The results from each of the methods were highly correlated (r > 0·95). Naturally contaminated processed soy products on the market were also studied. There were no discrepancies observed between the respective methods. Conclusions: TEMPO methods were equivalent to the corresponding standard plate methods with very good rates of agreement. Significance and Impact of the Study: The automated MPN method is more practical and reliable for in‐house microbiological testing in processed soy products.     

Screening soy hydrolysates for the production of a recombinant therapeutic protein in commercial cell line by combined approach of near-infrared spectroscopy and chemometrics

Soy hydrolysates are widely used as the major nutrient sources for cell culture processes for industrial manufacturing of therapeutic recombinant proteins. The primary goal of this study was to develop a spectroscopy based chemometric method, a partial least squares (PLS), to screen soy hydrolysates for better yield of protein production (titers) in cell culture medium. Harvest titer values of 29 soy hydrolysate lots with production yield between 490 and 1,350 mg/L were obtained from shake flask models or from manufacture engineering runs. The soy hydrolysate samples were measured by near-infrared (NIR) in reflectance mode using an infrared fiber optic probe. The fiber optic probe could easily enable in situ measurement of the soy hydrolysates for convenient raw material screening. The best PLS calibration has a determination coefficient of R ²?=?0.887 utilizing no spectral preprocessing, the two spectral ranges of 10,000–5,376 cm^?1 and 4,980–4,484 cm^?1, and a rank of 6 factors. The cross-validation of the model resulted in a determination coefficient of R ²?=?0.741 between the predicted and actual titer values with an average standard deviation of 72 mg/L. Compared with the resource demanding shake flask model, the combination of NIR and chemometric modeling provides a convenient method for soy hydrolysate screening with the advantage of fast speed, low cost and non-destructive.     

Susceptibility of psychrotrophic pseudomonads of milk origin to psychrotrophic bacteriophages.

A total of 47 psychrotrophic pseudomonads isolated from raw milk came from Newfoundland (19 isolates), British Columbia (6 isolates), Ontario (19 isolates), and Cork, Ireland (3 isolates). The susceptibility of these was tested against 30 bacteriophages isolated from cold-storage beef. Distinct lysotypes were observed with isolates representing different geographic regions. Phages as agents in the control of bacterial populations in milk is discussed.     

Diversity among lactococci isolated from ewes' raw milk and cheese

P. GAYA, M. BABÍN, M. MEDINA and M. NUÑEZ.1999.The technological and genetic characteristics of lactococci present in ewes' raw milk and 1-d-old ewes' raw milk cheeses sampled over a 1-year period were investigated. The proportion of lactic acid bacteria isolates from milk samples able to decrease milk pH by more than 1·25 units after 6 h incubation at 30 °C reached 14·5% in spring vs 10·7% in summer, 8·3% in autumn and 3·0% in winter. In 1-d-old cheese samples, the proportion of lactic acid bacteria able to lower milk pH by more than 1·25 units increased up to 32·3% in spring vs 23·4% in summer, 8·0% in autumn and 10·3% in winter. Fast acid-producing lactic acid bacteria mainly belonged to the genus Lactococcus . Using polymerase chain reaction protocols, fast acid-producing lactococci were grouped as 61  Lactococcus lactis subsp. lactis , 13  L. lactis subsp. cremoris and 14  L. lactis subsp. lactis biovar diacetylactis. Randomly amplified polymorphic DNA (RAPD) fingerprinting of fast acid-producing lactococci, using two primers, resulted in 21 different RAPD patterns for L. lactis subsp. lactis isolates, nine RAPD patterns for L. lactis subsp. cremoris isolates and three RAPD patterns for L. lactis subsp. lactis biovar diacetylactis isolates. Up to 19 different RAPD patterns were found for L. lactis isolates from cheeses made in a particular month.     

Contamination of milk by enterococci and coliforms from bovine faeces

AIM: To determine the contribution of enterococci and coliforms from bovine faeces and teats to contamination of raw milk. Methods: Putative enterococci (n = 301) and coliforms (n = 365) were isolated from bovine faeces (n = 20), cows' teats (n = 20), the raw milk (n = 1) and the milking environment (n = 4) on one farm. The clonal relationships of each bacterial group were investigated using Pulsed-Field Gel Electrophoresis of genomic macrorestriction fragments. Representatives of the different clusters of enterococci were identified by molecular techniques including rep-PCR, SDS protein profiling, Fluorescent Amplified Fragment Length Polymorphism (FAFLP), phenylalanyl-tRNA synthase (pheS) sequence analysis and/or 16S rDNA gene sequencing. Coliforms were identified by API 20E strips. RESULTS: The majority of the bovine faecal enterococcal isolates were identified as a potential new species of Aerococcus (100 isolates); E. faecium (28 isolates), and Aerococcus viridans (28 isolates) were also found. All coliform isolates from the bovine faeces were identified as Escherichia coli. The coliforms present in the milk were Hafnia alvei, Serratia liquefaciens, Yersinia enterocolitica and Enterobacter amnigenus. No E. coli, Enterococcus or Aerococcus from the bovine faeces were found in the milk. A single clone of H. alvei was found in the water, the milking equipment and the milk, suggesting that the water was the source of the organism in the milk. No vancomycin-resistant aerococci or enterococci were found while most of the isolates tested showed the presence of at least one virulence gene. The milk-sock retained strains that adhered to particulate faecal material. Coliforms were present at approx. 2 orders of magnitude greater than enterococci in the bovine faeces. CONCLUSIONS: The results imply that bovine faeces are not an important source of contamination of raw milk with enterococci or coliforms. SIGNIFICANCE AND IMPACT OF THE STUDY: The results confirm those of two previous studies (Gelsomino et al. 2001, Int J Food Microbiol71, 177-188 and Kagkli et al. 2007, Int J Food Microbiol114, 243-251) on two other farms. The three studies show that contamination of milk by enterococci, lactobacilli and coliforms of bovine faecal origin is extremely low. The results also suggest that where raw milk is implicated in food infection, other factors in addition to faecal contamination must be involved.     

An outbreak of staphylococcal food poisoning caused by enterotoxin H in mashed potato made with raw milk

Mashed potato made with raw bovine milk was suspected to have been the source of a food poisoning outbreak. Almost 8 x 10(8)Staphylococcus aureus CFU g(-1) were detected in the mashed potato. S. aureus was also found in bulk milk from the farm that had supplied milk for the mashed potato. Isolates from mashed potato and bulk milk were positive for the gene encoding staphylococcal enterotoxin H (seh), and the corresponding protein toxin, SEH, was detected by ELISA in the mashed potato. Production of SEH by S. aureus isolates from mashed potato (n = 4) and bulk milk (n = 4) was also demonstrated by ELISA. Sequencing of seh from one mashed potato isolate and one bulk milk isolate confirmed that the gene was a variant seh, and that the genes in both isolates were identical. Macrorestriction of chromosomal DNA with Sma1 followed by pulsed-field gel electrophoresis of seh-positive S. aureus from mashed potato and bulk milk revealed indistinguishable banding patterns between isolates from both sources. It seems likely that raw bovine milk used in the preparation of mashed potato contained S. aureus that subsequently produced sufficient SEH in the mashed potato to cause food poisoning.