摘要
利用傅里叶变换红外光谱仪采集6个不同地区的24个淡菜样品的醇提物红外光谱,原始光谱经过多点基线校正、聚类分析,采用平均值法建立淡菜的红外指纹图谱共有模式。相似度分析表明,各样品红外光谱与所建立的指纹图谱共有模式比较相似度良好,所选用建立共有模式的21批样品相似度均大于0.9,符合指纹图谱要求;而聚类分析中剔除的3个样品(S7、S10和S21)相似度均低于0.9,不符合指纹图谱研究要求,相似度分析和聚类分析结果相吻合。主成分分析建模结果表明,前两个主成分能代表原始数据96%的信息,相同产地的样品在主成分散点图上聚集为同一的类群,基本实现不同产地淡菜的鉴别。载荷因子分析结果表明,不同产地淡菜醇提物的差别主要体现在不饱和脂肪酸和磷脂的含量上。海口、厦门产区的淡菜样品所含不饱和脂肪酸含量较高;大连、烟台产区的淡菜样品含有较多的磷脂成分,舟山产区不饱和脂肪酸及磷脂含量均较低。因此,红外光谱指纹图谱结合聚类分析和主成分分析法可以快速、无损地鉴别不同产地淡菜,并且能反映不同产地淡菜醇提物中不饱和脂肪酸和磷脂含量的差异。
It was well known that aquatic products collected from different locations might considerably differ in their types and quantities of nutrient components, therefore, resulting in different quality. It is generally believed that these differences are essentially caused by various environmental conditions and habitats where the aquatic products are grown and harvested. Therefore, the quality control of aquatic products was a major concern for both the health authorities and the public. Hence, a validated method was essential for a quality assessment point of view. Because of its advantages and popularity, fingerprint anal- ysis was widely accepted and used in quality control systems of herbal medicines, and some methods have been proposed for the fingerprint analysis, including infrared spectrum （IR）, gas chromatography （GC）, high-pressure liquid chromatography （HPLC） and gas chromatography-mass spectrometry （GC-MS）; but fingerprint technology was rarely used in the quality control of aquatic products at present. For the first time, a validated infrared spectrum method coupled with cluster analysis, principal component analysis and load factor analysis methods had been developed for the study of the infrared fingerprint of mussel, and Fourier transform infrared （FTIR） spectrometer was performed on the 24 batches samples collected from Dalian, Yantai, Qingdao, Zhoushan, Xiamen and Haikou. Similarity analysis results indicated that similarity of 21 batches samples was higher than 0.9, which accords with the finger- print technique criterion. The samples from the same region had the similar characteristic peaks of infrared spectra and could be clustered together. The normalized spectra was selected to construct principal com- ponent analysis model in the range of fingerprint region 1 800-800 cm-1, and according to the model, the first two principal components （PC 1 and PC2） accounted for 96% of the variance information in the fin- gerprint region, and each sample was able to form distinct clust
出处
《水产学报》
CAS
CSCD
北大核心
2012年第7期1146-1152,共7页
Journal of Fisheries of China
基金
国家自然科学基金项目(81001393)
浙江省水产品加工创新团队项目(2009R50031-16)
舟山市科技计划项目(2011C22060)
关键词
淡菜
红外光谱
指纹图谱
相似度
主成分分析
载荷因子分析
mussel
infrared spectrum
fingerprint
similarity
principal component analysis
load factor analysis
